Compare commits
2 Commits
e79c19a929
...
e489dc4c27
Author | SHA1 | Date | |
---|---|---|---|
e489dc4c27 | |||
34149980a4 |
288
cmd/main.go
288
cmd/main.go
@@ -6,6 +6,7 @@ import (
|
|||||||
"fmt"
|
"fmt"
|
||||||
"io"
|
"io"
|
||||||
"os"
|
"os"
|
||||||
|
"slices"
|
||||||
|
|
||||||
"github.com/fatih/color"
|
"github.com/fatih/color"
|
||||||
|
|
||||||
@@ -25,6 +26,8 @@ func main() {
|
|||||||
multiLineFlag := flag.Bool("t", false, "Multi-line mode. Treats newline just like any character.")
|
multiLineFlag := flag.Bool("t", false, "Multi-line mode. Treats newline just like any character.")
|
||||||
printMatchesFlag := flag.Bool("p", false, "Prints start and end index of each match. Can only be used with '-t' for multi-line mode.")
|
printMatchesFlag := flag.Bool("p", false, "Prints start and end index of each match. Can only be used with '-t' for multi-line mode.")
|
||||||
caseInsensitiveFlag := flag.Bool("i", false, "Case-insensitive. Disregard the case of all characters.")
|
caseInsensitiveFlag := flag.Bool("i", false, "Case-insensitive. Disregard the case of all characters.")
|
||||||
|
recursiveFlag := flag.Bool("r", false, "Recursively search all files in the given directory.")
|
||||||
|
lineNumFlag := flag.Bool("n", false, "For each line with a match, print the line number. Implies -l.")
|
||||||
matchNum := flag.Int("m", 0, "Print the match with the given index. Eg. -m 3 prints the third match.")
|
matchNum := flag.Int("m", 0, "Print the match with the given index. Eg. -m 3 prints the third match.")
|
||||||
substituteText := flag.String("s", "", "Substitute the contents of each match with the given string. Overrides -o and -v")
|
substituteText := flag.String("s", "", "Substitute the contents of each match with the given string. Overrides -o and -v")
|
||||||
flag.Parse()
|
flag.Parse()
|
||||||
@@ -64,21 +67,30 @@ func main() {
|
|||||||
// 2. Build NFA from postfix representation (Thompson's algorithm)
|
// 2. Build NFA from postfix representation (Thompson's algorithm)
|
||||||
// 3. Run the string against the NFA
|
// 3. Run the string against the NFA
|
||||||
|
|
||||||
if len(flag.Args()) < 1 || len(flag.Args()) > 2 { // flag.Args() also strips out program name
|
if len(flag.Args()) < 1 { // flag.Args() also strips out program name
|
||||||
|
fmt.Println("ERROR: Missing cmdline args")
|
||||||
|
os.Exit(22)
|
||||||
|
}
|
||||||
|
if *recursiveFlag && len(flag.Args()) < 2 { // File/Directory must be provided with '-r'
|
||||||
fmt.Println("ERROR: Missing cmdline args")
|
fmt.Println("ERROR: Missing cmdline args")
|
||||||
os.Exit(22)
|
os.Exit(22)
|
||||||
}
|
}
|
||||||
var re string
|
var re string
|
||||||
re = flag.Args()[0]
|
re = flag.Args()[0]
|
||||||
var inputFile *os.File
|
var inputFiles []*os.File
|
||||||
if len(flag.Args()) == 1 || flag.Args()[1] == "-" { // Either no file argument, or file argument is "-"
|
if len(flag.Args()) == 1 || flag.Args()[1] == "-" { // Either no file argument, or file argument is "-"
|
||||||
inputFile = os.Stdin
|
if !slices.Contains(inputFiles, os.Stdin) {
|
||||||
|
inputFiles = append(inputFiles, os.Stdin) // os.Stdin cannot be entered more than once into the file list
|
||||||
|
}
|
||||||
} else {
|
} else {
|
||||||
var err error
|
inputFilenames := flag.Args()[1:]
|
||||||
inputFile, err = os.Open(flag.Args()[1])
|
for _, inputFilename := range inputFilenames {
|
||||||
if err != nil {
|
inputFile, err := os.Open(inputFilename)
|
||||||
fmt.Printf("%s: No such file or directory\n", flag.Args()[1])
|
if err != nil {
|
||||||
os.Exit(2)
|
fmt.Printf("%s: No such file or directory\n", flag.Args()[1])
|
||||||
|
os.Exit(2)
|
||||||
|
}
|
||||||
|
inputFiles = append(inputFiles, inputFile)
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
@@ -86,160 +98,166 @@ func main() {
|
|||||||
var err error
|
var err error
|
||||||
var linesRead bool // Whether or not we have read the lines in the file
|
var linesRead bool // Whether or not we have read the lines in the file
|
||||||
lineNum := 0 // Current line number
|
lineNum := 0 // Current line number
|
||||||
// Create reader for stdin and writer for stdout
|
// Create writer for stdout
|
||||||
reader := bufio.NewReader(inputFile)
|
|
||||||
out := bufio.NewWriter(os.Stdout)
|
out := bufio.NewWriter(os.Stdout)
|
||||||
|
// Compile regex
|
||||||
regComp, err := reg.Compile(re, flagsToCompile...)
|
regComp, err := reg.Compile(re, flagsToCompile...)
|
||||||
if err != nil {
|
if err != nil {
|
||||||
fmt.Println(err)
|
fmt.Println(err)
|
||||||
return
|
return
|
||||||
}
|
}
|
||||||
for true {
|
|
||||||
if linesRead {
|
for _, inputFile := range inputFiles {
|
||||||
break
|
reader := bufio.NewReader(inputFile)
|
||||||
}
|
linesRead = false
|
||||||
if !(*multiLineFlag) {
|
for true {
|
||||||
// Read every string from stdin until we encounter an error. If the error isn't EOF, panic.
|
if linesRead {
|
||||||
test_str, err = reader.ReadString('\n')
|
break
|
||||||
lineNum++
|
}
|
||||||
if err != nil {
|
if !(*multiLineFlag) {
|
||||||
|
// Read every string from stdin until we encounter an error. If the error isn't EOF, panic.
|
||||||
|
test_str, err = reader.ReadString('\n')
|
||||||
|
lineNum++
|
||||||
|
if err != nil {
|
||||||
|
if err == io.EOF {
|
||||||
|
linesRead = true
|
||||||
|
} else {
|
||||||
|
panic(err)
|
||||||
|
}
|
||||||
|
}
|
||||||
|
if len(test_str) > 0 && test_str[len(test_str)-1] == '\n' {
|
||||||
|
test_str = test_str[:len(test_str)-1]
|
||||||
|
}
|
||||||
|
} else {
|
||||||
|
// Multi-line mode - read every line of input into a temp. string.
|
||||||
|
// test_str will contain all lines of input (including newline characters)
|
||||||
|
// as one string.
|
||||||
|
var temp string
|
||||||
|
for temp, err = reader.ReadString('\n'); err == nil; temp, err = reader.ReadString('\n') {
|
||||||
|
test_str += temp
|
||||||
|
}
|
||||||
|
// Assuming err != nil
|
||||||
if err == io.EOF {
|
if err == io.EOF {
|
||||||
|
if len(temp) > 0 {
|
||||||
|
test_str += temp // Add the last line (if it is non-empty)
|
||||||
|
}
|
||||||
linesRead = true
|
linesRead = true
|
||||||
} else {
|
} else {
|
||||||
panic(err)
|
panic(err)
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
if len(test_str) > 0 && test_str[len(test_str)-1] == '\n' {
|
matchIndices := make([]reg.Match, 0)
|
||||||
test_str = test_str[:len(test_str)-1]
|
if matchNumFlagEnabled {
|
||||||
}
|
tmp, err := regComp.FindNthMatch(test_str, *matchNum)
|
||||||
} else {
|
if err == nil {
|
||||||
// Multi-line mode - read every line of input into a temp. string.
|
matchIndices = append(matchIndices, tmp)
|
||||||
// test_str will contain all lines of input (including newline characters)
|
|
||||||
// as one string.
|
|
||||||
var temp string
|
|
||||||
for temp, err = reader.ReadString('\n'); err == nil; temp, err = reader.ReadString('\n') {
|
|
||||||
test_str += temp
|
|
||||||
}
|
|
||||||
// Assuming err != nil
|
|
||||||
if err == io.EOF {
|
|
||||||
if len(temp) > 0 {
|
|
||||||
test_str += temp // Add the last line (if it is non-empty)
|
|
||||||
}
|
}
|
||||||
linesRead = true
|
|
||||||
} else {
|
} else {
|
||||||
panic(err)
|
matchIndices = regComp.FindAllSubmatch(test_str)
|
||||||
}
|
}
|
||||||
}
|
|
||||||
matchIndices := make([]reg.Match, 0)
|
|
||||||
if matchNumFlagEnabled {
|
|
||||||
tmp, err := regComp.FindNthMatch(test_str, *matchNum)
|
|
||||||
if err == nil {
|
|
||||||
matchIndices = append(matchIndices, tmp)
|
|
||||||
}
|
|
||||||
} else {
|
|
||||||
matchIndices = regComp.FindAllSubmatch(test_str)
|
|
||||||
}
|
|
||||||
|
|
||||||
test_str_runes := []rune(test_str) // Converting to runes preserves unicode characters
|
test_str_runes := []rune(test_str) // Converting to runes preserves unicode characters
|
||||||
|
|
||||||
if *printMatchesFlag {
|
if *printMatchesFlag {
|
||||||
// if we are in single line mode, print the line on which
|
// if we are in single line mode, print the line on which
|
||||||
// the matches occur
|
// the matches occur
|
||||||
if len(matchIndices) > 0 {
|
if len(matchIndices) > 0 {
|
||||||
if !(*multiLineFlag) {
|
if !(*multiLineFlag) {
|
||||||
fmt.Fprintf(out, "Line %d:\n", lineNum)
|
fmt.Fprintf(out, "Line %d:\n", lineNum)
|
||||||
}
|
}
|
||||||
for _, m := range matchIndices {
|
for _, m := range matchIndices {
|
||||||
fmt.Fprintf(out, "%s\n", m.String())
|
fmt.Fprintf(out, "%s\n", m.String())
|
||||||
}
|
}
|
||||||
err := out.Flush()
|
err := out.Flush()
|
||||||
if err != nil {
|
if err != nil {
|
||||||
panic(err)
|
panic(err)
|
||||||
}
|
|
||||||
}
|
|
||||||
continue
|
|
||||||
}
|
|
||||||
// Decompose the array of matchIndex structs into a flat unique array of ints - if matchIndex is {4,7}, flat array will contain 4,5,6
|
|
||||||
// This should make checking O(1) instead of O(n)
|
|
||||||
indicesToPrint := new_uniq_arr[int]()
|
|
||||||
for _, idx := range matchIndices {
|
|
||||||
indicesToPrint.add(genRange(idx[0].StartIdx, idx[0].EndIdx)...)
|
|
||||||
}
|
|
||||||
// If we are inverting, then we should print the indices which _didn't_ match
|
|
||||||
// in color.
|
|
||||||
if *invertFlag {
|
|
||||||
oldIndices := indicesToPrint.values()
|
|
||||||
indicesToPrint = new_uniq_arr[int]()
|
|
||||||
// Explanation:
|
|
||||||
// Find all numbers from 0 to len(test_str_runes) that are NOT in oldIndices.
|
|
||||||
// These are the values we want to print, now that we have inverted the match.
|
|
||||||
// Re-initialize indicesToPrint and add all of these values to it.
|
|
||||||
indicesToPrint.add(setDifference(genRange(0, len(test_str_runes)), oldIndices)...)
|
|
||||||
|
|
||||||
}
|
|
||||||
// If lineFlag is enabled, we should only print something if:
|
|
||||||
// a. We are not inverting, and have at least one match on the current line
|
|
||||||
// OR
|
|
||||||
// b. We are inverting, and have no matches at all on the current line.
|
|
||||||
// This checks for the inverse, and continues if it is true.
|
|
||||||
if *lineFlag {
|
|
||||||
if !(*invertFlag) && len(matchIndices) == 0 || *invertFlag && len(matchIndices) > 0 {
|
|
||||||
continue
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
// If we are substituting, we need a different behavior, as follows:
|
|
||||||
// For every character in the test string:
|
|
||||||
// 1. Check if the index is the start of any matchIndex
|
|
||||||
// 2. If so, print the substitute text, and set our index to
|
|
||||||
// the corresponding end index.
|
|
||||||
// 3. If not, just print the character.
|
|
||||||
if substituteFlagEnabled {
|
|
||||||
for i := range test_str_runes {
|
|
||||||
inMatchIndex := false
|
|
||||||
for _, m := range matchIndices {
|
|
||||||
if i == m[0].StartIdx {
|
|
||||||
fmt.Fprintf(out, "%s", *substituteText)
|
|
||||||
i = m[0].EndIdx
|
|
||||||
inMatchIndex = true
|
|
||||||
break
|
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
if !inMatchIndex {
|
continue
|
||||||
fmt.Fprintf(out, "%c", test_str_runes[i])
|
}
|
||||||
|
// Decompose the array of matchIndex structs into a flat unique array of ints - if matchIndex is {4,7}, flat array will contain 4,5,6
|
||||||
|
// This should make checking O(1) instead of O(n)
|
||||||
|
indicesToPrint := new_uniq_arr[int]()
|
||||||
|
for _, idx := range matchIndices {
|
||||||
|
indicesToPrint.add(genRange(idx[0].StartIdx, idx[0].EndIdx)...)
|
||||||
|
}
|
||||||
|
// If we are inverting, then we should print the indices which _didn't_ match
|
||||||
|
// in color.
|
||||||
|
if *invertFlag {
|
||||||
|
oldIndices := indicesToPrint.values()
|
||||||
|
indicesToPrint = new_uniq_arr[int]()
|
||||||
|
// Explanation:
|
||||||
|
// Find all numbers from 0 to len(test_str_runes) that are NOT in oldIndices.
|
||||||
|
// These are the values we want to print, now that we have inverted the match.
|
||||||
|
// Re-initialize indicesToPrint and add all of these values to it.
|
||||||
|
indicesToPrint.add(setDifference(genRange(0, len(test_str_runes)), oldIndices)...)
|
||||||
|
|
||||||
|
}
|
||||||
|
// If lineFlag is enabled, we should only print something if:
|
||||||
|
// a. We are not inverting, and have at least one match on the current line
|
||||||
|
// OR
|
||||||
|
// b. We are inverting, and have no matches at all on the current line.
|
||||||
|
// This checks for the inverse, and continues if it is true.
|
||||||
|
if *lineFlag {
|
||||||
|
if !(*invertFlag) && len(matchIndices) == 0 || *invertFlag && len(matchIndices) > 0 {
|
||||||
|
continue
|
||||||
|
} else {
|
||||||
|
color.New(color.FgMagenta).Fprintf(out, "%s: ", inputFile.Name()) // Print filename
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
} else {
|
|
||||||
for i, c := range test_str_runes {
|
// If we are substituting, we need a different behavior, as follows:
|
||||||
if indicesToPrint.contains(i) {
|
// For every character in the test string:
|
||||||
color.New(color.FgRed).Fprintf(out, "%c", c)
|
// 1. Check if the index is the start of any matchIndex
|
||||||
// Newline after every match - only if -o is enabled and -v is disabled.
|
// 2. If so, print the substitute text, and set our index to
|
||||||
if *onlyFlag && !(*invertFlag) {
|
// the corresponding end index.
|
||||||
for matchIdxNum, idx := range matchIndices {
|
// 3. If not, just print the character.
|
||||||
if matchIdxNum < len(matchIndices)-1 { // Only print a newline afte printing a match, if there are multiple matches on the line, and we aren't on the last one. This is because the newline that gets added at the end will take care of that.
|
if substituteFlagEnabled {
|
||||||
if i+1 == idx[0].EndIdx { // End index is one more than last index of match
|
for i := range test_str_runes {
|
||||||
fmt.Fprintf(out, "\n")
|
inMatchIndex := false
|
||||||
break
|
for _, m := range matchIndices {
|
||||||
|
if i == m[0].StartIdx {
|
||||||
|
fmt.Fprintf(out, "%s", *substituteText)
|
||||||
|
i = m[0].EndIdx
|
||||||
|
inMatchIndex = true
|
||||||
|
break
|
||||||
|
}
|
||||||
|
}
|
||||||
|
if !inMatchIndex {
|
||||||
|
fmt.Fprintf(out, "%c", test_str_runes[i])
|
||||||
|
}
|
||||||
|
}
|
||||||
|
} else {
|
||||||
|
for i, c := range test_str_runes {
|
||||||
|
if indicesToPrint.contains(i) {
|
||||||
|
color.New(color.FgRed, color.Bold).Fprintf(out, "%c", c)
|
||||||
|
// Newline after every match - only if -o is enabled and -v is disabled.
|
||||||
|
if *onlyFlag && !(*invertFlag) {
|
||||||
|
for matchIdxNum, idx := range matchIndices {
|
||||||
|
if matchIdxNum < len(matchIndices)-1 { // Only print a newline afte printing a match, if there are multiple matches on the line, and we aren't on the last one. This is because the newline that gets added at the end will take care of that.
|
||||||
|
if i+1 == idx[0].EndIdx { // End index is one more than last index of match
|
||||||
|
fmt.Fprintf(out, "\n")
|
||||||
|
break
|
||||||
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
} else {
|
||||||
} else {
|
if !(*onlyFlag) {
|
||||||
if !(*onlyFlag) {
|
fmt.Fprintf(out, "%c", c)
|
||||||
fmt.Fprintf(out, "%c", c)
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
err = out.Flush()
|
||||||
err = out.Flush()
|
if err != nil {
|
||||||
if err != nil {
|
panic(err)
|
||||||
panic(err)
|
}
|
||||||
}
|
// If the last character in the string wasn't a newline, AND we either have don't -o set or we do (and we've matched something), then print a newline
|
||||||
// If the last character in the string wasn't a newline, AND we either have don't -o set or we do (and we've matched something), then print a newline
|
if (len(test_str_runes) > 0 && test_str_runes[len(test_str_runes)-1] != '\n') &&
|
||||||
if (len(test_str_runes) > 0 && test_str_runes[len(test_str_runes)-1] != '\n') &&
|
(!*onlyFlag || indicesToPrint.len() > 0) {
|
||||||
(!*onlyFlag || indicesToPrint.len() > 0) {
|
fmt.Println()
|
||||||
fmt.Println()
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
Reference in New Issue
Block a user