diff.go

package main

import (
	"bufio"
	"fmt"
	"os/exec"
	"strconv"
	"strings"
)

type PrecedingLineMatch struct {
	Content           string
	Filename          string
	LineNumber        int
	ContentSimilarity float32
}

const LINE_ADD = 0
const LINE_UPDATE = 5
const LINE_MOVE = 10
const LINE_MOVE_UPDATE = 15

const MINIMUM_RAW_SIMILARITY = 0.74
const MINIMUM_SIMILARITY_FIRST_TERM_ONLY = 0.9
const MINIMUM_SIMILARITY_SUBSEQUENT_TERMS = 0.8

const STATE_FREE_FORM = 0
const STATE_IN_DIFF = 10

func FindPredecessorLineFromDiff(directory string, sha string, targetFilename string, targetLineNumber int, lineContents string) LineChange {
	cmd := exec.Command("git", "rev-list", "--parents", sha)
	cmd.Dir = directory
	stdout, err := cmd.StdoutPipe()
	if err != nil {
		FailAndExit("Error getting parent list for %s: %v", sha, err)
	}
	defer stdout.Close()
	err = cmd.Start()
	if err != nil {
		FailAndExit("Error starting git rev-list command: %v", err)
	}

	scanner := bufio.NewScanner(stdout)
	scanner.Scan()
	line := scanner.Text()
	revListParents := strings.Split(line, " ")
	if len(revListParents) > 1 {
		// We can use the tilde some_sha~ syntax to mean "parent of some_sha"
		shaWithTilde := fmt.Sprintf("%s~", sha)
		cmd = exec.Command("git", "diff", shaWithTilde, sha)
	} else {
		// no parents = we need to diff vs an explicit "blank repo" sha
		// From https://stackoverflow.com/questions/40883798/how-to-get-git-diff-of-the-first-commit
		BlankRepoSha := "4b825dc642cb6eb9a060e54bf8d69288fbee4904"
		cmd = exec.Command("git", "diff", BlankRepoSha, sha)
	}
	cmd.Dir = directory

	stdout, err = cmd.StdoutPipe()
	if err != nil {
		FailAndExit("Error creating diff for %s: %v", sha, err)
	}
	defer stdout.Close()

	err = cmd.Start()
	if err != nil {
		FailAndExit("Error starting git diff command for %s: %v", sha, err)
	}

	scanner = bufio.NewScanner(stdout)
	currentFilename := ""
	currentLineNumber := 0
	previousLineNumber := 0
	maxLineNumber := 0
	state := STATE_FREE_FORM
	var precedingLineMatch *PrecedingLineMatch = nil
	overallLineNumber := 0

	for scanner.Scan() {
		overallLineNumber++
		line := scanner.Text()

		if state == STATE_FREE_FORM && strings.HasPrefix(line, "diff --git a/") {
			splitLine := strings.Split(line, " ")
			fileBit := splitLine[2]
			currentFilename = fileBit[2:]
			currentLineNumber = 0
		} else if state == STATE_FREE_FORM && strings.HasPrefix(line, "@@ ") {
			splitLine := strings.Split(line, " ")
			fromLines := strings.Split(splitLine[1], ",")
			toLines := strings.Split(splitLine[2], ",")
			previousLineNumber, err = strconv.Atoi(fromLines[0])
			previousLineNumber = absint(previousLineNumber)
			if err != nil {
				FailAndExit("Error converting old offset: %v", err)
			}
			var currentLineNumber int
			var changeCount int
			if len(toLines) > 1 {
				currentLineNumber, err = strconv.Atoi(toLines[0])
				if err != nil {
					FailAndExit("Error converting old offset: %v", err)
				}
				changeCount, err = strconv.Atoi(toLines[1])
				if err != nil {
					FailAndExit("Error converting diff line count: %v", err)
				}
			} else {
				currentLineNumber = 0
				changeCount, err = strconv.Atoi(toLines[0])
				if err != nil {
					FailAndExit("Error converting diff line count: %v", err)
				}
			}
			maxLineNumber = currentLineNumber + changeCount - 1
			state = STATE_IN_DIFF
		} else if state == STATE_IN_DIFF {
			if strings.HasPrefix(line, "-") {
				contentSimilarity := LineContentSimilarity(lineContents, line[1:])
				if contentSimilarity > MINIMUM_RAW_SIMILARITY {
					if precedingLineMatch == nil || precedingLineMatch.ContentSimilarity >= contentSimilarity {
						precedingLineMatch = &PrecedingLineMatch{
							Content:           line[1:],
							Filename:          currentFilename,
							LineNumber:        previousLineNumber,
							ContentSimilarity: contentSimilarity,
						}
					}
				}
				previousLineNumber++
			} else if strings.HasPrefix(line, "+") {
				currentLineNumber++
			} else {
				if strings.HasPrefix(line, " ") && line[1:] == lineContents && (precedingLineMatch == nil || precedingLineMatch.ContentSimilarity < 1) {
					// we found a context line that matches, so treat it as overriding *unless we also found a previous line that exactly matches*
					precedingLineMatch = &PrecedingLineMatch{
						Content:           line[1:],
						Filename:          currentFilename,
						LineNumber:        previousLineNumber,
						ContentSimilarity: 1.0,
					}
				}
				currentLineNumber++
				previousLineNumber++
			}
			if currentLineNumber >= maxLineNumber {
				state = STATE_FREE_FORM
			}
		}
	}

	if err := scanner.Err(); err != nil {
		FailAndExit("Error reading stdout: %v", err)
	}

	err = cmd.Wait()
	if err != nil {
		FailAndExit("Got error waiting for diff %s to finish: %v", sha, err)
	}

	if precedingLineMatch == nil {
		return LineChange{
			Sha:        sha,
			ChangeType: LINE_ADD,
			Filename:   targetFilename,
			LineNumber: targetLineNumber,
			Content:    lineContents,
			Previous:   nil,
			Commit:     DetailsForCommit(directory, sha),
		}
	} else {
		var changeType int
		if precedingLineMatch.LineNumber != targetLineNumber && precedingLineMatch.Content != lineContents {
			changeType = LINE_MOVE_UPDATE
		} else if precedingLineMatch.Content != lineContents {
			changeType = LINE_UPDATE
		} else {
			changeType = LINE_MOVE
		}
		return LineChange{
			Sha:        sha,
			ChangeType: changeType,
			Filename:   targetFilename,
			LineNumber: targetLineNumber,
			Content:    lineContents,
			Previous:   BlameAndDeriveChange(directory, &sha, &precedingLineMatch.Content, precedingLineMatch.Filename, precedingLineMatch.LineNumber),
			Commit:     DetailsForCommit(directory, sha),
		}
	}
}