US 4,510,567 AGrant
Qualifying and Sorting File Record Data
Issue Date:1985-04-09
•8 Claims
•6 Drawing Sheets
Abstract
Record data on a disk file is sorted in a text/data processor by means of an algorithm that transfers such records on the basis of rank to a sort buffer on the basis of qualifying criteria. Each qualified file record is compared with the lowest ranked record previously transferred and located in the sort buffer. When a higher ranked record is identified it is transferred into the buffer at a location based on qualification. Lower ranked records are deleted from the sort buffer if space does not permit the storing of such records within the space available. When the sort buffer has been loaded with the highest ranked records remaining in the disk file without overflowing the buffer is unloaded to an output device. The sort program recycles through a subsequent pass again transferring the highest ranked remaining records into the sort buffer. To minimize recycle time, a presort algorithm is run to set record identifying bits in a bit map section of the sort buffer. Each time the contents of the sort buffer is output the record identifying bits for the records in the sort buffer are reset to a second state. The second state of a record identifying bit indicates that that record will not be considered for future passes through the sort routine.
Metadata
Assignee
- International Business Machines Corp.
Inventors
- Philip Y. Chang
- John W. McInroy
Application Information
Application Number:US 2647974
Filing Date:1981-05-18
Priority Date:1981-05-18
Art Unit:237
Classifications
IPC:
G06F 700G06F 724
Field of Search:
364200 MS File;900 MS File
Patent Drawings (6 sheets)
Description
Technical Field
This invention relates to file record data handling in a text/data processing system, and more particularly to file record data sorting in a limited memory environment.
Background Art
Office machines for processing text data to produce hard copies are being utilized with increasing frequency for data processing. Such text processing machines are physically structured to include all the necessary elements for data processing including a display screen for presenting to an operator a visual presentation of record data as it will appear on hard copy or be utilized by an output device. Typically, text processing machines incorporate a conventional typewriter keyboard with additional control keys for operator interface with the system controls. For data processing, operator instructions are input through the keyboard and displayed on a screen in a manner similar to the use of such machines for text processing. The display is segmented to allow datad formatting instructions to be presented to the operator along with a visual presentation of the data records being processed. Thus, the dispaly not only provides a visual presentation to the operator of the record data being processed, but also provides prompting instructions to allow interface with a text processing machine as a data operator.
Text processing machines on which data processing is carried out include electronic hardware programmed to evaluate operator keystrokes and to control the display. This electronic hardware includes memory for storing file format information and instruction data for future use for processing the file records for subsequent utilization or preparation of hard copy. Responding to the keystrokes of an operator are system controls including programmed computers which prepare the display screen for presenting to the operator file record data. Ease of operation of the machine for the operator is in large part determined by the electronics and associated computers and programs therefor.
Data processing on a text processing machine is achievable by utilization of algorithms with the machines's computers and memory. There are, however, several differences between the processing of text data and record data in a text processing machine. Typically, text is stored in an unstructured, but formatted, format while file records are stored in a structured, but unformatted manner. Text data is formatted in that it includes tabs, carrier returns, etc. Due to the fact that a data file can be fairly large, structuring of the included data permits rapid access to any particular portion. However, this requires programming to achieve this rapid access. To assist an operator in the accessing of file record etc, the system display screen in segmented into virtual displays or frames that includes grids to be filled in by an operator for processing of the file data. Through keying, an operator can then cause the stored record data to be qualified for keying "search" criteria.
File record data processing in text processing machines usually utilizes storage of the file record data on a disk. In the past, sorting of the data on the disk was carried out externally, that is, data was transferred from one disk to a second disk in a sort operation or sorting was completed in internal memory. With some earlier file record data processing in a text processing machine, a disk access is needed for the sorting of each record which is too cumbersome and time consuming.
Sorting of data is one of the most utilized functions of a text processing machine as a data processor and it thus becomes important to provide efficiency and ease of operator interface of the sorting function.
Disclosure of the Invention
According to the present invention, data sorting is achieved by use of a sort buffer for temporarily storing qualified file records. Typically, the sort buffer may be a random access memory having, for example 40 K-bytes of storage. Only qualified records are stored in the sort buffer after being qualified on the basis of one of more search fields. The process of the present invention minimizes the need for a large amount of memory space sufficient to store an entire file, or as an alternative, to use a disk work space to store intermediate results.
To provide file data sorting on the basis of qualifying fields, file data (for example, as stored on the disk) is sequentially loaded into a sort buffer based on specified qualifications and sort criteria. The file data is transferred from the disk to the sort buffer until the buffer cannot accept the next qualified record. However, the file data in th sort buffer is not utilized at this time but instead, the remainder of the file is read and evaluated for comparison against file data previously transferred to the sort buffer. During this reading operation, records of a higher ranking replace records of a lower ranking previously transferred to the sort buffer. Upon completion of a reading of all of the file data, a register is set to identify the lower ranked record in the sort buffer and the contents of the buffer are stored or printed. The buffer is cleared for accepting additional records.
The above operation is repeated beginning with the lowest ranked record in the previous operation and continuing until the sorting of the record data file is complete. That is, for each pass the sort buffer is loaded with the best available record file data and the contents are stored or printed. The operation resumes at the lower ranked record previously transferred to the buffer.
In addition to removing the requirement for disk work space, the record data file procesing of the present invention enables the utilization of different size sort buffers. Where high speed sorting is required to sort buffer with greater storage space will provide such faster sorting. The larger the sort buffer, the fewer passes that would be required througha file to sort each record on a selected equalification criteria. Thus, a large buffer enables fast sorting and a buffer of more restricted space will provide corresponding slower sorting. An operator of a text processing machine has some control over the speed of sorting by utilization of extra sort buffer space with the process of the present invention.
In accordance with the present invention, there is provided a method of sorting file record data in a storage means for temporary retention in a sort buffer in an order based on selected ranking criteria; the method comprises the steps of: (a) ranking each file record in the storage means as to an order of preferential arrangement in the sort buffer; (b) next the process compares a file record in the storage means on the basis of ranking criteria with the lowest ranked file record previously transferred to the sort buffer. A compared file record in the storage means is transferred to unused space in the sort buffer and is located at a level based on ranking, that is, a higher ranked record will displace downward in the bufferd a lower ranked record.
When the sort buffer has been previously loaded to capacity, a higher ranked file record in the storage means will be transferred into the sort buffer and a lower ranked record will be deleted. Again, the file record transferred to the sort buffer will be located at a position depending on its ranking. When the sort buffer has been filled with the highest ranked file records in the storage means still available for sorting, then a register is set to identify the lowest ranked record in the buffer. The buffer is then cleared for storage or printing. The process recycles starting at the record having a ranking one position lower than the last record in the buffer previously transferred for storage or printing.
Brief Description of the Drawings
A more complete understanding of the invention and its advantages will be apparent from the specification and claims and from the accompanying drawings illustrative of the invention.
Referring to the drawings:
FIG. 1 is a basic block diagram of a text/data processing system including a keyboard, processor, memory storage and a display screen for visual presentation to an operator of text data and operating or prompting instructions;
FIG. 2 is a bock diagram showing logic hardware detail of the processor of FIG. 1 with interconnecting control and data bus paths;
FIG. 3 is a detail showing the formatting of the storage of the system of FIG. 1 as structured for data processing;
FIG. 4 is a block diagram showing data transfer and structuring of a sort buffer for receiving qualified data records;
FIG. 5 is a high level flow chart for the basic algorithm for processing record data from a disk to the sort buffer on a record qualified basis;
FIG. 6 is a more detailed flow chart of the basic algorithm as generalized in FIG. 5 showing the major functions for sorting file record data;
FIG. 7 is a detailed flow chart for qualifying, transferring and locating record data on a disk storage to a sort buffer for datad handling in the text/data processing system of FIG. 1; and
FIG. 8 is a flow chart of an algorithm for processing record data prior to sorting and transferring to a sort buffer as detailed by the flow chart of FIG. 7.
Detailed Description
Referring to FIG. 1, a text and data processing system is shown including a processor 10, interconnected by means of a data bus 12 and control lines 14 to a keyboard 16. Included on the keyboard are conventional alpha-numeric characters found on the standard typewriter keyboard. In addition the keyboard includes symbol keys, such as backspace, carrier return and tab settings along with function keys used to instruct the system to perform certain operations, to control the way a display is created, or to provide information about stored textual and/or record data. In addition, the keyboard includes code keys that are used in conjunction with another key to input special instructions to the processor 10.
Keystroke data transferred to the processor 10 over the data bus 12 in the form of 7-bit bytes is processed by instructions stored in a random access memory 18 by means of an inteconnecting data bus 20. The random access memory 18 stores the keystroked data bytes, and file record data in addition to instructions for operation of the processor 10.
Also connected to the random access memory 18 by means of a data bus 22 is a display refresh buffer 24 that receives display data for formatting a visual presentation on a display screen 26. Control signals for operation of the display screen 26 ar provided along a data bus 28.
Typically, the display screen 26 is a cathode ray tube, in a preferred embodiment of the present invention, was formatted into multiple virtual images, the major one of which displayed twenty lines of text or data information. The total line capacity of the display was twenty-five lines with the first line reserved as a virtual display for document formatting information and the bottom two lines of the display reserved for virtual images as a prompt line and a message line to give an operator instructions on the control of the text/data processing system of FIG. 1.
Referring to FIG. 2, the processor 10 is further detailed to show typical logic hardware elements as found in such processors. The processor may be a commercially available unit, such as from Intel Corporation and identified by the Number 8086. Typically the processor includes a control logic unit 30 which responds to interrupts on a device bus 32 from the keyboard 16. The control logic unit 30 is also connected to the datad and address bus 34 interconnected to various other logic units of the processor 10.
In response to a fetch instruction from the random access memory, the control logic unit 30 generates control signals to other logic elements of the processor. These control signals are interconnected to the various elemens by means of a control line 36 which is illustrated directly connected to an arithmetic logic unit 38 and identified as a "control" line 36 to other elements of the processor. Synchronous operation of the control unit 30 with other logic elements of the processor 10 is achieved by means of clock pulses input to the processor from an external clock source. This instruction bus is also shown interconnected to othe logic elements of the processor detailed on FIG. 2.
Data and instructions to be processed in the processor 10 are input through a bus control logic unit 42. Data to be processed may also come from program input/output control logic 44. The bus control logic 42 interconnects storage elements of the random access memory 18 and receives instructions for processing data received from the input/output control 44 or received from the random access memory. Thus, the input/output control 44 receives data from the keyboard 16 or the random access memory 18 while the bus control logic 42 receives instructions and/or data from the same memory. Note that different storage sections of the random access memory are identifiable for program storage and data storage.
Device control information from the processor 10 is output through the program input/output controller 44 over a data bus (12, 20). Input data on the data bud (12, 20) from the keyboard 16 is processed internally through the processor by instructions on the bus 34 to the status register 60. The arithmetic logic unit 38 in response to a control signal on line 36 and in accordance with instructions received on the memory bus 46 performs arithmatic computations which may be stored in temporary scratch registers 52. Various other transfers of data betwen the arithmatic logic unit 38 and other logic elements of the processor are of course possible. Such additional transfers may be to a status register 60, a data point register 56 or a stack pointer register 58. Also in the data stream for these various logic elements by means of the bus 34 is a program counter 54.
A particular operating sequence for the processor 10 is determined by instructions and data on the memory bus 46 and input data on the bi-directional bus (12, 20). As an example, in response to received instructions, the processor transfers data stored in the scratch registers 52 to one of the registers 56, 58 or 60. Such operations of processors as detailed in FIG. 2 are considered to be well known and understood by one of ordinary skill in the data processing field. A detailed description of each operation of the processor of FIG. 2 for the described invention would be counterproductive to an understanding of the invention as claimed.
Referring to FIG. 3, there is shown a block diagram of a typical operational format of the system of FIG. 1 for data processing and includes the keyboard 16 connected by way of a data bus 12 to the record display buffer 24. Also connected to the record display buffer 24 is the display 26 by means of the data bus 28. As formatted, the data processing system includes a printer 70 and a disk drive 80, all external devices, connected to data storage bufers enclosed within the dotted outline. It should be understood that each of the identifiable areas of buffer storage within the dotted outline comprise storage locations addressable by conventional techniques. Thus, keystrokes from the keyboard 16 are received by the display buffer 24 over the data bus 12.
Stored within memory of the system are application programs consisting of instructions and data bytes transferred between the various buffers including the display buffer 24, a sort buffer 64, a text storage buffer 76, and a data input/output buffer 82. As the same implies, the next storage buffer 76 contains a data stream of data for processing in accordance with the present invention.
The text storage buffer interconnects by means of data lines to the printer 70 and the disk drive 80. Record data from the disk drive 80 is transferred through the data input/output buffer 82 into the sort buffer 64 as will be explained. Records properly sorted within the sort buffer 64 are transferred through the input/output buffer 82 through the display buffer 24 to the display 26 by means of the data line 28. This record data is also transferred through the text storage buffer 76 to the printer 70 where stored instructions direct the making of a permanent hard copy record.
By operation of the system of FIGS. 1-3 in conjunction with the processor 10, in the manner to be described with reference to FIGS. 4-18, a sorting of file record data is made for presentation on the display screen 26.
Referring to FIG. 4, records to be sorted are stored on a disk which is inserted by the disk drive 80 for processing in accordance with a sort program previously stored. Stored on the disk along with a data record is the byte length of the record which information is required for sorting in accordance with the sorting program. Data records on the disk are sorted by evaluating one or more fields such as last name, zip code, and age. Typically, up to five fields are available for sorting data records. Each life record is read from the disk by the disk drive 80 and evaluated by the sort program and transferred through the input/output buffer 82 for storage in the sort buffer 64.
As configured, the sort buffer 64 includes a record header section 86, a record data section 88, and bit map section 90, the latter being explained with reference to FIG. 8. The record heater section includes a list of the length of each record in the data section 88 along with a pointer identifying the location of the record in the data section. While each entry into the header section 86 is of a fixed length, entries into the record data section vary with the length of the record transferred from the disk drive 80. In one embodiment on the invention, the sort program stores the record header bytes starting from the first portion of the record section 86 and locates the record data starting at the last position in the record section 88. This technique makes maximum use of the sort buffer size.
The speed with which sorting of record data from the disk drive 80 is completed depends on the storage length of the sort buffer 64. The sort algorithm to be detailed with reference to FIGS. 5-7 is a multiple pass algorithm that continues to transfer record data into the data section 88 until the next stored record overflows the datad section. The records sorted and stored in the data section 88 are then transferred out of the sort buffer 64 through the input/output buffer 82 to the display 26, or through the text storage buffer 76 to the printer 70, or for other use as appropriate. Thus, the larger the sort buffer, the fewer passes required of the sort program of clearing all records on a disk. Typically, the sort buffer varies from 8 K-bytes in length to 40 K-bytes which includes the three sections, 86, 88 and 90.
Referring to FIG. 5, after the processor 10 receives a signal that a disk has been inserted into the disk drive 80, the sort program starts at entry 92 and proceeds to run the sequence of steps to complete the process of block 94. That is, each data record on the disk is qualified for selection and ranked for sorting the arranging in the sort buffer 64. The first record accessed on the disk is transferred through the input/output buffer 82 to the first position in the data section 88. The header section 86 is also loaded with pointers identifying the record, the record length in bytes and the address of the record in the data section 88.
After this first record has been located in the data section 88, a test 96 is made by the sort program to determine if there are additional records on the disk that have not been sorted and processed to the data section 88. If additional records are remaining on the disk, then the program recycles to block 94 to evaluate the next record on the disk. Ranking criteria of the next record is compared with ranking criteria of the previously transferred record to determine if it has a higher ranking or a lower ranking. For a record having a lower ranking than the previously transferred record, the data is transferred through the buffer 82 and located in the data section 88 at the next available storage location. Information for the next record is stored in the header section 86 which, as previously described, includes both record length and record location in the data section.
the algorithm recycles through 94 and 96 so long as there is a record on the disk that has not previously been transferred to the buffer 64. When the next record on the disk has ranking criteria with a higher level of ranking than previously transferred records, the next record will be transferred through the buffer 82 and located in the data section 88 below the next highest ranked record in the data section. All records in the data section 88 having a lower ranking are relocated in the data section. Thus, each time a record is located in the data section 88 having a higher ranking than previously transferred records, a reassembly of the data is required. For example, if the records are being sorted on last names, and records A, B, F, M, and R have been previously transferred to the data section 88 and the next record is sorted on the record D, then all records having a ranking lower than D will be shifted in the data section 88. The new arrangement will then be as follows: A, B, D, F, M, R.
The above description of the sorting process assumes that space remains in the data section 88 for all of the previously transferred records. For each transfer of a record into the sort buffer 64, an evaluation is made by the sort program to determine if additional byte space is available in the data section for storge of transferred records. If the next record to be transferred has a lower ranking then any previously transferred records and its length would cause in overflow of the data section 88, then this next record will not be transferred into the data section. It should be noted, however, that the sort program continues to evaluate each record on the disk to determine if one of higher ranking than previously transferred records remains. A record of a higher ranking, as explained, will replace a record of a lower rank. If the transferring of a record having a higher ranking causes the data section to overflow, then records of a lower rank will be deleted.
After the data section 88 has been filled to the extent possible without overflowing with the highest ranked records not previously sorted from the disk, the sort program makes the test 96 which results in an affirmative response and the program advances to a "use record" step 98 which outputs the record in the data section 88 to the disk 80 and also outputs the records to the printer 70 for producing hard copy. The data may also be used in other processes as required.
Referring to FIG. 6, there is shown in greater detail the sort algorithm as previously outlined in FIG. 5. The sort algorithm starts at entry 92 and completes a test 100 to determine if all the records on the disk have been processed for possible transfer into the data section 88. If the end of file test produces a negative result, then the algorithm advances to a step 102 to access the next record on the disk not previously sorted and transferred to the buffer 64 without subsequent deletion. The next record is evaluated at test 104 to determine if it qualifies for the sort program that is being run. That is, if the sorting is being made on last names beginning with the letters A through M and the next record is a last name beginning with the letter R, then the next record does not qualify for the sort and the sequence returns to the test 100. So long as the additional records are to be sorted from the disk the loop including 100, 102 and 104 recycles until a qualified record is found at test 104.
Upon location of a qualified record on the disk a test 106 is made to determine if the next record has a higher level of ranking than the last record located in the buffer 64. When the next record has a lower ranking than the last record located in the buffer then the test 106 produces a negative result and the sequence advances to a test 108 to determine if the next record will fit in the buffer 64 without overflowing the data section 88. If the length of the next record is in excess of the space available in the data section 88, then the sort sequence returns to the test 100.
The sequence including 100, 102, 104, 106 and 108 recycles each time the next qualified record has a lower level or ranking than the last record in the buffer 64 and the length of this record is greater than the space available in the data section 88.
When space is available in the data section 88 as determined by the test 108, then the sequence advances to a step 110 to transfer the next record through the buffer 82 into the data section 88. This next record is located following the last record stored in the data section. The identification, length and location of this record is also stored in the header section 86. After this next record has been added to the data section 88 at sequence step 110, the sort algorithm recycles to the "end of file" test 100.
When the next record has a higher ranking than the lowest ranked record in the buffer 64, the test 106 produces an affirmative result which advances the sequence to a test 112 to determine if sufficient byte space remains in the buffer 64 to accept the record under consideration. If the test 112 produces a negative response, the sequence advances to a step 114 which deletes the lowest ranked record from the buffer 64 and the processor recycles to the test 112. The sequence of test 112 and the step 114 are recycled, each time deleting the lowest ranked record from the buffer 64, until an affirmative result is produced at the test 112.
An affirmative result from the test 112 advances the sequence to a step 116 and the record which resulted in the affirmative result of the test 106 is added to the buffer 64 in the data section 88 with the record located in the sort order as determined by the sort algorithm. Following the location of this next record to the sort buffer 64 the sequence recycles to the "end of file" test 100.
The sequence of tests and steps 100, 102, 104, 106, 108, 110, 112, 114 and 116 is run until all the file records have been evaluated and the data section 88 has no more available byte space to accept the next qualified record on the disk. At this time the "end of file" test 100 produces an affirmative result and the sequence advances to a test 118. If any records are found to have been located in the data section 88 the test 118 produces an affirmative result advancing the sequence to a step 120 to configure the input/output buffer 82 to transfer the data record over the input/output bus to the disk 80 or other output device including the hard copy printer 70.
At the completion of the transfer of the records in the sort buffer 64 to the disk 80 or other output device, the sequence returns to test 100. The sequence starting at test 100 and continuing through step 116 is repeated for the next highest ranked records on the disk, that is, the next highest ranked following the lowest ranked record previously transferred from the buffer 64. The process as described above is recycled again loading the data section 88 with the highest ranked records still remaining on the disk. The test 118 and step 120 are again run and this next batch of records are transferred to the disk 80 or other output device. This recycle process continues until the test 118 produces a negative result indicating that on the previous pass no records were added to the data section 88 signifying that all records on the disk have been sorted. The sequence then steps to an "end" step 122 and the sort program shuts down the processor.
For a more detailed explanation of the sort process, reference is now made to FIG. 7. The flow chart of FIG. 7 adds additional detail to the flow chart of FIG. 6 and only such additional detail will be described with reference to FIG. 7. The sequence starts at entry 92 and after the test 100 the sequence advances to the "access next record" step 102 which is followed by a test to determine the condition of the input/output buffer 82. This test is made at 124. This sequence of the algorithm minimizes data movement between the disk drive 80 and the sort buffer 64 by loading the buffer 82 prior to transfer of data into the sort buffer. That is, file records from the disk drive 80 are batched into the input/output buffer 82 until this buffer is full. With the buffer 82 full, or not able to accept the next record, the test 124 produces an affirmative result and the sequence advances to a step 126 which moves all the data in the input/output buffer 82 into the sort buffer 64 as a batch.
The header section 86 is continuously updated with new pointers for the additional records being input and located into the data section 88. Note, that only those records having new pointers in header section 86 are transferred to data section 88. Thus, not all records transferred into the buffer 82 are subsequently transferred to the section 88. Those records not transferred from the buffer 82 are deleted from the buffer 82.
After the data in the input/output buffer 82 has been transferred into the sort buffer 64 the test 124 produces a negative result of the test 104 advances the sequence to a step 128 which compares the sort fields of the next qualifying record on the disk with the sort fields of the lowest ranked record in the data section 88. It is during the running of this step that the location of the record is determined when transferred to the data section 88. After the level of ranking has been determined at the step 128, the sequence advances to the test 106 followed by the test 108 which evaluates if the free space left in the sort buffer 64 is greater than the length of the next qualifying record. As previously explained, a negative result recycles the sequence and a positive result adds the record at step 110 which also causes the sequence to recycle.
A positive result of the test 106 advances the sequence to the test 112 which determines if the free space left in the buffer 64 is greater than the length of the new record. If not, the step 114 is run to delete the last record to increase free space in the data section 88. When enough free space is made available in the data section 88 to accept the next qualifying record on the disk, the sequence advances through an affirmative route to the step 116 which makes a binary search of the data in the header section 86 to evaluate where the next record fits into the records previously transferred and located in the data section 88. Data identifying the record, its length, and its loction in the data section 88, are inserted into the header 86 during running of the step 116. The sequence then returns to the test 100. When the data section 88 can accept no more records from the disk the sequence advances through the affirmative route to the test 118 and the step 120 or the end 122.
Referring to FIG. 8, there is shown a flow chart of an expansion of the basic sort algorithm of FIGS. 5-7 which provides increased sort efficiency. The algorithm of FIG. 8 can be considered a presort sequence and utilizes the bit map section 90 of the sort buffer 64. With the program of FIG. 8 loaded into the processor 10 along with the basic sort program of FIGS. 5-7, the sorting of records on a disk inserted into the disk drive 80 commences at a start 130. Note, where the flow chart of FIG. 8 incorportes tests or steps previously described the same reference number will be applied.
Following the start 130, a "first pass" test 132 is made to determine if the records on the disk have been previously qualified by a first pass algorithm. In the first pass through the file the records are qualified and the test proceeds along the affirmative path to the first pass algorithm 134 which is detailed in FIG. 7 and commences at the "access next record" step 102. The algorithm of FIG. 7 is completed for the first pass to identify qualified records on the disk. Note, that in order to select the qualified records all the records in the file must be evaluated and qualified. At the same time, a record identifying bit is set for each record in the bit map section 90 by sequencing the algorithm of FIG. 8 through the test 136 along the negative path back to the first pass test 132.
When all the record identifying bits in the bit map section 90 are set to a first state (a logic ONE) and the data section 88 no longer contains sufficient space to accept the next highest ranked record, the test 100 produces an affirmative result advancing the sequence through the test 118 and the "use record" step 120 or step to the end 122.
Upon completion of the first pass all the record identifying bits of the records in the data section 88 that have been read out for use in the disk 80, or other output device, will have been set to the first state and then reset to the second state (logic ZERO) indicating that the record has been sorted and need not be considered again for future passes through the disk file. All other records on the disk will have the record identifying bit set to the first state. Upon completion of the first pass the test 132 causes the sequence to proceed along the negative path to a test 138 to determine if additional first state bits are found in the section 90. If additional first state bits are available the negative path from the test 138 is followed to an "access record" step 140 which searches the disk file for the next record with the record identifying bit set to one, that is, the first state. Following identification of the next record with an identifying bit set to one, the test 124 is made which proceeds along the affirmative path to the "move record" step 126 and back to the "end of bit map" test 138.
If additional space remains in the buffer 64 the sequence of FIG. 8 advances to an "insert record" step 142 which begins at the comparison step 128 of FIG. 7 and follows the path there detailed. However, the negative path from the test 108 returns the sequence of FIG. 8 to the "end of bit map" test 138 as does a completion of the "add record" step 110 and also a completion of the binary search step 116. Thus, the "insert record" step 142 includes comparision of the header data in the header section 86 and space management of the sort buffer 64 as previously described. Note, that during the pass of the sequence through the step 140 all those records that have the record identifying bit set to zero, that is, the second state, will be bypassed thereby reducing the number of records that must be evaluated for transfer to the data section 88 of the buffer 64.
Following the running of step 142, each time the data section 88 includes the most qualified records remaining in the disk file having a record identifying bit in the first state, the sequence including test 118 and step 120 are run to output these records as previously described. The records thus transferred from the data section 88 will have the record identifying bit then set to zero, that is, the second state.
The sequence through path of test 138 and step 140 continues until all of the record identifying bits in the bit map section 90 are reset to the second state as shown by a positive result of the test 138. Following the positive path from the test 138 the sequence advances to a bit overflow test 144 which produces a negative result when the bit map section 90 had sufficient space to contain a record identifying bit for each record on the disk. Thus, the sequence from test 138 through step 140 continuesfor each bit in the bit map until the last bit has been checked.
With the last bit in the bit map section 90 checked and a negative result from the test 144 the sequence advances to the test 136 and if a record identifying bit in the map section 90 is in the first state, that is, a one, then the program recycles to the first pass test 132 and again takes up at the "end of bit map" test 138. This sequence continues until all the record identifying bits have been set to zero advancing the program through the test 136 along the affirmative path to the end 146. At this time all the qualifying records on the disk have been sorted through the sort buffer 64 and output to the disk 80 or other output device.
When the number of records to be sorted is greater than the space available for record identifying bits in the map section 90, the first pass algorithm 134 will be run as previously described with one exception. That is, when the last record identifying bit is located, and more than one record remains on the disk, all remaining records will be assigned to the last record identifying bit. After the first pass the test 132 advances the sequence to the test 138 which inserts records into the data section 88 as previously explained through step 140, test 142, step 126 and step 142.
When all of the record identifying bits in the bit map section have been reset to the second state as determined by the test 138 the overflow test 144 is made which, when the overflow bit has been set, will produce a positive result advancing the presort algorithm to a step 146. Step 146 resets the processor to run the program of FIG. 7 for each record covered by the overflow bit. The algorithm of FIG. 7 is run until the "end of file" test 100 produces a positive response thereby advancing the sequence to the test 136 which recycles the algorithm of FIG. 8 to the first pass test 132.
After all the records assigned to the overflow bit have been evaluated for sorting and the data section 88 loaded with the highest ranked records the algorithm follows through the path of tests 132 and 138, through the step 140, again sorting records assigned identifying bits. Again the test 144, and the step 146 are completed for records in the overflow bit. The presort program then continues until all the records in the disk file have been sorted into the buffer 64 and transferred to an output device.
Each of the tests and steps of the algorithms as illustrated by the flow charts of FIGS. 7 and 8 is also described by the psuedocode of the following table. The psuedocode is a list of each instruction for running the sort and presort algorithms on the test/data processing machine of FIG. 1.
While only one embodiment of the invention, together with modifications thereof, has been described in detail herein and shown in the accompanying drawings, it will be evident that various further modifications are possible without departing from the scope of the invention.
Claims
We claim:
1. A method of qualifying and sorting file records from a mass storage means into a sort buffer memory for use in an office processing environment whereby the sort buffer memory provides a temporary retention of the file records in an order based on a selected ranking, comprising the steps of: (a) accessing each of said file records in said mass storage means; (b) evaluating an accessed file record in accordance with a chosen sort algorithm to determine if it qualifies for transfer to the sort buffer memory; (c) if the accessed file record qualifies, then comparing a rank of the qualified file record in the mass storage means with the rank of a lowest ranked file record in the sort buffer memory; (d) if the rank of the qualified file record is higher than the rank of the lowest ranked file record in the sort buffer memory, then the qualified file record is transferred to unused space in the sort buffer memory; and if the rank of the qualified record is lower than the rank of the lowest ranked file record, then the qualified file record is not transferred to the sort buffer memory if the buffer is full; (e) placing the qualified file record transferred to the sort buffer memory at a level based on the rank determined by the chosen sorting algorithm; (f) repeating steps (c)--(e) for each qualified file record in the storage means until the sort buffer has been loaded with the highest ranked records in the storage means without overflowing; and (g) outputting said highest ranked records from said sort buffer.
2. A method of qualifying and sorting file records as set forth in claim 1 wherein the step of evaluating an accessed file record to determine if it qualifies for transfer includes evaluating multiple fields of each file record.
3. A method of qualifying and sorting file records as set forth in claim 1 including the step of repeating steps (c)-(g) for the remaining qualified file records in the storage means.
4. A method of qualifying and sorting file records as set forth in claim 3 wherein the file records in the sort buffer are transferred to a disk storage.
5. A method of qualifying and sorting file records in a storage means into a sort buffer for temporary retention in an order based on a selected ranking, comprising th steps of: (a) accessing each of said file records in said storage means; (b) evaluating an accessed file record to determine if it qualifies for transfer to the sort buffer; (c) if said accessed file record qualifies, comparing a rank of the qualified file record in the storage means with the rank of a lowest ranked file record in the sort buffer; (d) if the rank of the qualified file record is higher than the rank of the lowest ranked file record, transferring the qualified file record in the storage means to the sort buffer; (e) deleting the lowest ranked file record from the sort buffer to facilitate storage of the qualified file record therein; (f) locating the qualified file record transferred to the sort buffer at a level based on the rank thereof; and (g) repeating steps (c)-(f) for each qualified file record in the storage means until the sort buffer has been loaded with the highest ranked records without overflowing.
6. A method of qualifying and sorting file records as set forth in claim 5 including the step of outputting the file records in the sort buffer after steps (c)-(g) have been repeated for each file record in the storage means for the highest ranked file records in the storage means not previously transferred to the sort buffer.
7. A method of qualifying and sorting file records as set forth in claim 6 including the step of repeating the outputting of file records from the sort buffer each time said buffer contains the highest ranked file records from the storage means not previously transferred to the sort buffer.
8. A method of qualifying and sorting file records as set forth in claim 7 including the step of transferring file records in the sort buffer to a disk storage.
Patent Citations (7)
Non-Patent Literature (4)
- Software Tools in Pascal, Kernigan and Plauges published by Bell Labs. Inc. 1981 --Chapter 4.
- An Operator Orientated Data Base Management System by J. Shapiro published in Kilobaud, Mar. 1980, p. 84.
- Data Base Management by K. Widelitz published in Kilobaud, May 1979, p. 54.
- Sorting Routines by Andrew J. Rerko, Kilobaud, Apr. 1977, pp. 34-36.