diff options
Diffstat (limited to 'main.go')
| -rw-r--r-- | main.go | 236 |
1 files changed, 236 insertions, 0 deletions
@@ -0,0 +1,236 @@ +package main + +// Takes one file as the `src` and another as the `check` (passed via -c). +// Ensures that all entries in `check` are in `src`, returning as missing all +// entries in `check` that are missing from `src`. +// To check all that are in `src` missing in `check`, you can firstly simply +// switch around the order of parameters, or otherwise use bt-present, which +// returns either a true or false that all entries on `src` are on `check`, +// which could be used to weed out any incorrect values on `src`. + +import ( + bt "badtudexo" + st "saggytrousers" + xl "github.com/xuri/excelize/v2" + "os" + "fmt" +) + +func usage() { + fmt.Println( +`bt-missing: check whether entries on check are missing on src. + +You must pass two files: a source file (src, passed as an argument) and +a check file (check, passed with --check). This script will iterate +over all entries in the check file, and output all which are missing on +the source file. + +This script has fundmentally two uses. + +The first is: given a larger file, check all which are _not_ on the +smaller file. To perform this, run: + bt-missing -c large.xlsx small.xlsx + +The second is: expand the output of bt-present. bt-present returns a +boolean whether all entries on the source file are present in the +check file. If bt-present returns false, you may wish to get the +entries that are missing. To do this, invoke bt-missing with the +positions of the arguments reversed: + bt-present -c check.xlsx src.xlsx + bt-missing -c src.xlsx check.xlsx +This can be a little confusing - if in doubt, just switch them around +and see whichever one looks right! + +You may specify the sheet and the columns to check against as flags +here so you do not need to specify them when the script is running. + +You can additionally specify an output file with -o or --output; the +output of this command will be copied into the output file. + +If the specific sheets and columns are known, it can help to pass them +as flags, saving you to have to confirm them when the program is +running. + +Flags: +-h, --help Display this help message. +-c, --check The file to check against; you must pass this. +-ss, --ssheet The sheet to check against (for src). +-cs, --csheet The sheet to check against (for check). +-sc, --scolumn The column to check against (for src) as a string (i.e. not an index) +-cc, --ccolumn The column to check against (for check) as a string (i.e. not an index) +-o, --output The path of an output file to store the output in. +-q, --quiet Be quiet: do not return the results to stdout. + This means this script does not show if there are any + missing unless you also pass -o. +`) + os.Exit(0) +} + +func parseArgs(args []string) { + var check bool = false + + for i := 0; i < len(args); i++ { + arg := args[i] + switch arg { + case "-h", "--help", "help": + usage() + case "-c", "--check", "--ck": + i++ + g_ckfile = args[i] + check = true + case "-cs", "--csheet", "--check-sheet", "--ck-sheet": + i++ + g_cksheet = args[i] + case "-cc", "--ccolumn", "--check-column", "--ck-column": + i++ + g_ckcol = args[i] + case "-ss", "--ssheet", "--source-sheet", "--src-sheet": + i++ + g_srcsheet = args[i] + case "-sc", "--scolumn", "--source-column", "--src-column": + i++ + g_srccol = args[i] + case "-o", "--out", "--output": + i++ + g_out = args[i] + case "-q", "--quiet": // Do not print to stdout. + quiet = true + + default: + g_srcfile = arg + } + } + + if !check { + fmt.Println("You need to pass --check.") + os.Exit(-1) + } + + if quiet && g_out == "" { + fmt.Println("You must pass --output if you pass -quiet.") + os.Exit(-1) + } +} + +// Global Variables +var g_ckfile string +var g_cksheet string +var g_ckcol string +var g_srcfile string +var g_srcsheet string +var g_srccol string +var g_out string +var quiet bool = false + +func main() { + args := os.Args + parseArgs(args) + +/* Open the check and source files. */ + // Load check file. + ck, err := xl.OpenFile(g_ckfile) + if err != nil { + fmt.Println("Could not open check file.") + return + } + // Load src file. + src, err := xl.OpenFile(g_srcfile) + if err != nil { + fmt.Println("Could not open src file.") + return + } + +/* Select the sheet and the column to operate on for the two files. */ + fmt.Println("Load the check file (you may be asked questions here.)") + cksheet := st.SelectSheet(ck, g_cksheet) + ckcol := st.SelectHeader(ck, cksheet, g_ckcol, /* exact: */ false) + + + fmt.Println("------ ------") + fmt.Println("Load the source file now (you may be asked questions here.)") + fmt.Println("------ ------") + srcsheet := st.SelectSheet(src, g_srcsheet) + srccol := st.SelectHeader(src, srcsheet, g_srccol, /* exact: */ false) + +/* Grab the columns of each files which are to be processed with. */ + ckrows, err := ck.GetCols(cksheet) + if err != nil { + fmt.Printf("Could not get the rows of sheet %s of the checkfile.\n", g_cksheet) + return + } + srcrows, err := src.GetCols(srcsheet) + if err != nil { + fmt.Printf("Could not get the rows of sheet %s of the sourcefile.\n", g_srcsheet) + return + } + ckrow := ckrows[ckcol] + srcrow := srcrows[srccol] + + // Output what is checked against what. + + + // Actually, like, do the checks. + missing := bt.Missing(ckrow, srcrow) + + // Write the output to stdout and to a file if -o passed. + printOutput(missing, g_out, cksheet, srcsheet, ckcol, srccol) +} + +func writeBoth(quiet, write bool, file *os.File, str, errstr string) { + if !quiet { fmt.Printf(str) } + if write { + _, err := file.WriteString(str) + if err != nil { + fmt.Printf("Could not write to file %d", errstr) + } + } +} + +func printOutput[T comparable](missing []bt.MissingValue[T], outputFile string, + cksheet, srcsheet string, + ckcol, srccol int) { + +/* Determine whether to write to an --out file and open it if we are. */ + writeFile := false + out, err := os.Create(outputFile) + if outputFile != "" && err != nil { + fmt.Println("Cannot open file to output.") + } else { + writeFile = true + } + +/* The data which was checked against. */ + writeBoth(quiet, writeFile, out, "--- Check File Information ---\n", "check file info") + writeBoth(quiet, writeFile, out, fmt.Sprintf("Filename: %s\n", g_ckfile), "check filename") + writeBoth(quiet, writeFile, out, fmt.Sprintf("Sheet: %s\n", cksheet), "check sheet") + writeBoth(quiet, writeFile, out, fmt.Sprintf("Column: %d\n", ckcol), "check column") + + writeBoth(quiet, writeFile, out, "--- Source File Information ---\n", "source file info") + writeBoth(quiet, writeFile, out, fmt.Sprintf("Filename: %s\n", g_srcfile), "check filename") + writeBoth(quiet, writeFile, out, fmt.Sprintf("Sheet: %s\n", srcsheet), "source sheet") + writeBoth(quiet, writeFile, out, fmt.Sprintf("Column: %d\n", srccol), "source column") + + +/* Write all missing entries. */ + for _, m := range missing { + str := fmt.Sprintf("Missing: \"%s\" at index %d.\n", m.Value, m.Index) + if !quiet { fmt.Printf(str) } + if writeFile { + _, err := out.WriteString(str) + if err != nil { + fmt.Printf("Cannot write missing entry %d to file\n", m.Index) + } + } + } + +/* Write total missing entries. */ + str := fmt.Sprintf("Total Missing: %d\n", len(missing)) + if !quiet { fmt.Printf(str) } + if writeFile { + _, err := out.WriteString(str) + if err != nil { + fmt.Printf("Could not write total missing to file.") + } + } + +} |