48c764841fe955d738b56e352082ca23dce91c96 galt Fri Mar 5 13:11:33 2010 -0800 added a copyright and my name and MarkDs name diff --git git-reports.c git-reports.c index 48e7dc4..178b3e1 100644 --- git-reports.c +++ git-reports.c @@ -1,800 +1,807 @@ -/* git-reports.c for creating git equivalent of cvs-reports. */ +/* git-reports.c for creating git equivalent of cvs-reports. + * + * Copyright Galt Barber 2010. + * + * Anyone is free to use it. + * Just include me in your credits. + * Likewise cvs-reports was created by Mark Diekhans. + */ #include "common.h" #include "options.h" #include "dystring.h" #include "errabort.h" #include "hash.h" #include "linefile.h" #include "htmshell.h" #include "portable.h" static char const rcsid[] = "$Id: git-reports.c,v 1.1 2010/03/02 08:43:07 galt Exp $"; struct hash *userHash = NULL; struct slName *users = NULL; char *startTag = NULL; char *endTag = NULL; char *startDate = NULL; char *endDate = NULL; char *title = NULL; char *repoDir = NULL; char *outDir = NULL; char *outPrefix = NULL; char gitCmd[1024]; char *tempMakeDiffName = NULL; struct files { struct files *next; char type; char *path; int linesChanged; }; struct commit { struct commit *next; int commitNumber; // used for sorting fileviews char *commitId; char *author; char *date; char *comment; struct files *files; }; struct comFile { struct comFile *next; struct files *f; struct commit *commit; }; void usage(char *msg) /* Explain usage and exit. */ { errAbort( "%s\n\n" "git-reports - produce source code reports useful for code-review on git repository \n" "\n" "Usage:\n" " git-reports startTag endTag startDate endDate title repoDir outDir outPrefix\n" "where " " startTag and endTag are repository tags marking the beginning and end of the git range\n" " startDate and endDate and title are just strings that get printed on the report\n" " title is usually the branch number, e.g. v225\n" " repoDir is where the git repository (use absolute path)\n" " outDir is the output directory (use absolute path).\n" " outPrefix is typically \"branch\" or \"review\" directory.\n" " --help - this help screen\n", msg); } static struct optionSpec options[] = { {"-help", OPTION_BOOLEAN}, {NULL, 0}, }; struct commit* getCommits() /* Get all commits from startTag to endTag */ { int numCommits = 0; safef(gitCmd,sizeof(gitCmd), "" "git log origin/%s..origin/%s --name-status > commits.tmp" , startTag, endTag); system(gitCmd); // TODO error handling struct lineFile *lf = lineFileOpen("commits.tmp", TRUE); int lineSize; char *line; struct commit *commits = NULL, *commit = NULL; struct files *files = NULL, *f = NULL; char *sep = ""; while (lineFileNext(lf, &line, &lineSize)) { boolean isMerge = FALSE; char *w = nextWord(&line); AllocVar(commit); if (!sameString("commit", w)) errAbort("expected keyword commit parsing commits.tmp\n"); commit->commitId = cloneString(nextWord(&line)); commit->commitNumber = ++numCommits; lineFileNext(lf, &line, &lineSize); w = nextWord(&line); if (sameString("Merge:", w)) { isMerge = TRUE; lineFileNext(lf, &line, &lineSize); w = nextWord(&line); } if (!sameString("Author:", w)) errAbort("expected keyword Author: parsing commits.tmp\n"); commit->author = cloneString(nextWord(&line)); lineFileNext(lf, &line, &lineSize); w = nextWord(&line); if (!sameString("Date:", w)) errAbort("expected keyword Date: parsing commits.tmp\n"); commit->date = cloneString(line); lineFileNext(lf, &line, &lineSize); if (!sameString("", line)) errAbort("expected blank line parsing commits.tmp\n"); /* collect the comment-lines */ struct dyString *dy = NULL; dy = dyStringNew(0); sep = ""; files = NULL; while (lineFileNext(lf, &line, &lineSize)) { if (sameString("", line)) break; w = skipLeadingSpaces(line); dyStringPrintf(dy, "%s%s", w, sep); sep = "\n"; } commit->comment = cloneString(dy->string); freeDyString(&dy); if (!isMerge) { /* collect the files-list */ while (lineFileNext(lf, &line, &lineSize)) { if (sameString("", line)) break; AllocVar(f); w = nextWord(&line); f->type = w[0]; f->path = cloneString(line); slAddHead(&files, f); } slReverse(&files); } commit->files = files; slAddHead(&commits, commit); verbose(2, "commitId: %s\n" "author: %s\n" "date: %s\n" "comment: [%s]\n" "file(s): \n" , commit->commitId , commit->author , commit->date , commit->comment); for (f=commit->files; f; f = f->next) { verbose(2, "%c %s\n", f->type, f->path); // anything other than M? if (f->type != 'M') verbose(2, "special type: %c %s\n", f->type, f->path); } verbose(2, "------------\n"); } lineFileClose(&lf); slReverse(&commits); unlink("commits.tmp"); return commits; } int makeHtml(char *diffPath, char *htmlPath, char *path, char *commitId) /* Make a color-coded html diff * Return the number of lines changed */ { int linesChanged = 0; FILE *h = mustOpen(htmlPath, "w"); struct lineFile *lf = lineFileOpen(diffPath, TRUE); int lineSize; char *line; char *xline = NULL; boolean inBody = FALSE; boolean inBlock = TRUE; int blockP = 0, blockN = 0; fprintf(h, "\n\n%s %s\n\n\n
\n", path, commitId);
 while (lineFileNext(lf, &line, &lineSize))
     {
     xline = htmlEncode(line);	
     if (line[0] == '-')
 	{
 	fprintf(h, "%s\n", xline);
 	if (inBody)
 	    {
 	    inBlock = TRUE;
 	    ++blockN;
 	    }
 	}
     else if (line[0] == '+')
 	{
 	fprintf(h, "%s\n", xline);
 	if (inBody)
 	    {
 	    inBlock = TRUE;
 	    ++blockP;
 	    }
 	}
     else
 	{
 	if (line[0] == '@')
 	    fprintf(h, "%s\n", xline);
 	else
 	    fprintf(h, "%s\n", xline);
 	if (inBody)
 	    {
 	    if (inBlock)
 		{
 		inBlock = FALSE;
 		if (blockP >= blockN)
 		    linesChanged += blockP;
 		else
 		    linesChanged += blockN;
 		blockP = 0;
 		blockN = 0;
 		}
 	    }
 	}
 	
     if (line[0] == '@')
 	inBody = TRUE;
 
     freeMem(xline);
 
     }
 // what if there is no last trailing line to end the last block?
 if (inBody)
     {
     if (inBlock)
 	{
 	inBlock = FALSE;
 	if (blockP >= blockN)
 	    linesChanged += blockP;
 	else
 	    linesChanged += blockN;
 	blockP = 0;
 	blockN = 0;
 	}
     }
 
 lineFileClose(&lf);
 fprintf(h, "
\n\n\n"); fclose(h); return linesChanged; } void makeDiffAndSplit(struct commit *c, char *u, boolean full) /* Generate a full diff and then split it up into its parts. * This was motivated because no other way to show deleted files * since they are not in repo and git paths must actually exist * in working repo dir. However leaving off the path produces * a diff with everything we want, we just have to split it up. */ { safef(gitCmd,sizeof(gitCmd), "git diff -b -w --no-prefix%s %s^ %s > %s" // TODO get proper temp file name , full ? " --unified=10000" : "" // should be good enough for most files , c->commitId, c->commitId, tempMakeDiffName); uglyf("gitCmd=%s\n", gitCmd); system(gitCmd); // TODO error handling // now parse it and split it into separate files with the right path. struct lineFile *lf = lineFileOpen(tempMakeDiffName, TRUE); int lineSize; char *line; FILE *h = NULL; while (lineFileNext(lf, &line, &lineSize)) { if (startsWith("diff --git ", line)) { if (h) { fclose(h); h = NULL; } //uglyf("line=%s\n", line); // DELETE THIS LINE char *fpath = line + strlen("diff --git "); fpath = strchr(fpath, ' '); ++fpath; // now we should be pointing to the world //uglyf("fpath=%s\n", fpath); char path[1024]; char *r = strrchr(fpath, '/'); if (r) { *r = 0; /* make internal levels of subdirs */ safef(path, sizeof(path), "mkdir -p %s/%s/%s/%s/%s/%s", outDir, outPrefix, "user", u, full ? "full" : "context", fpath); uglyf("path=%s\n", path); system(path); *r = '/'; } safef(path, sizeof(path), "%s/%s/%s/%s/%s/%s%s.diff" , outDir, outPrefix, "user", u, full ? "full" : "context", fpath, c->commitId); h = mustOpen(path, "w"); fprintf(h, "%s\n", c->commitId); fprintf(h, "%s\n", c->author); fprintf(h, "%s\n", c->date); fprintf(h, "%s\n", c->comment); } else if (startsWith("@@", line)) { char *end = strchr(line+2, '@'); *(end+2) = 0; // chop the weird unwanted context string from here following e.g. //@@ -99,7 +99,9 @@ echo // converts to //@@ -99,7 +99,9 @@ // saves 17 seconds over the more expensive sed command } fprintf(h, "%s\n", line); } if (h) { fclose(h); h = NULL; } lineFileClose(&lf); } void doUserCommits(char *u, struct commit *commits, int *saveUlc, int *saveUfc) /* process one user, commit-view */ { char userPath[1024]; safef(userPath, sizeof(userPath), "%s/%s/%s/%s/index.html", outDir, outPrefix, "user", u); FILE *h = mustOpen(userPath, "w"); fprintf(h, "\n\nCommits for %s\n\n\n", u); fprintf(h, "

Commits for %s

\n", u); fprintf(h, "switch to files view, user index\n"); fprintf(h, "

%s to %s (%s to %s) %s

\n", startTag, endTag, startDate, endDate, title); fprintf(h, "
\n");
 
 
 int userLinesChanged = 0;
 int userFileCount = 0;   // TODO do we want to not count the same file twice?
 
 char *cDiff = NULL, *cHtml = NULL, *fDiff = NULL, *fHtml = NULL;
 char *relativePath = NULL;
 char *commonPath = NULL;
 
 struct commit *c = NULL;
 struct files *f = NULL;
 for(c = commits; c; c = c->next)
     {
     if (sameString(c->author, u))
 	{
 	fprintf(h, "%s\n", c->commitId);
 	fprintf(h, "%s\n", c->date);
 	fprintf(h, "%s\n", c->comment);
 	makeDiffAndSplit(c, u, FALSE);
 	makeDiffAndSplit(c, u, TRUE);
 	for(f = c->files; f; f = f->next)
 	    {
 	    char path[1024];
 
             // context unified
 	    safef(path, sizeof(path), "%s/%s%s", "context", f->path, c->commitId);
 	    relativePath = cloneString(path);
 
 	    safef(path, sizeof(path), "%s/%s/%s/%s/%s", outDir, outPrefix, "user", u, relativePath);
 	    commonPath = cloneString(path);
 
 	    safef(path, sizeof(path), "%s.html", commonPath);
 	    cHtml = cloneString(path);
 
 	    safef(path, sizeof(path), "%s.diff", commonPath);
 	    cDiff = cloneString(path);
 
 	    // make context html page
 	    f->linesChanged = makeHtml(cDiff, cHtml, f->path, c->commitId);
 
 	    userLinesChanged += f->linesChanged;
 	    ++userFileCount;   // TODO do we want to not count the same file twice?
 
 	    freeMem(cDiff);
 	    freeMem(cHtml);
 	    safef(path, sizeof(path), "%s.html", relativePath);
 	    cHtml = cloneString(path);
 	    safef(path, sizeof(path), "%s.diff", relativePath);
 	    cDiff = cloneString(path);
 
 
 
             // full text (up to 10,000 lines)
 	    freeMem(relativePath);
 	    safef(path, sizeof(path), "%s/%s%s", "full", f->path, c->commitId);
 	    relativePath = cloneString(path);
 
 	    safef(path, sizeof(path), "%s/%s/%s/%s/%s", outDir, outPrefix, "user", u, relativePath);
 	    freeMem(commonPath);
 	    commonPath = cloneString(path);
 
 	    safef(path, sizeof(path), "%s.html", commonPath);
 	    fHtml = cloneString(path);
 
 	    safef(path, sizeof(path), "%s.diff", commonPath);
 	    fDiff = cloneString(path);
 
 
 	    //git show --unified=10000 11a20b6cd113d75d84549eb642b7f2ac7a2594fe src/utils/qa/weeklybld/buildEnv.csh
 
 	    // make full html page
 	    makeHtml(fDiff, fHtml, f->path, c->commitId);
 
 	    freeMem(fDiff);
 	    freeMem(fHtml);
 	    safef(path, sizeof(path), "%s.html", relativePath);
 	    fHtml = cloneString(path);
 	    safef(path, sizeof(path), "%s.diff", relativePath);
 	    fDiff = cloneString(path);
 
 	    // make file view links
 	    fprintf(h, "  %s - lines changed %d, "
 		"context: html, text, "
 		"full: html, text\n"
 		, f->path, f->linesChanged
 		, cHtml, cDiff, fHtml, fDiff);
 
 	    freeMem(relativePath);
 	    freeMem(commonPath);
 	    freeMem(cDiff);
 	    freeMem(cHtml);
 	    freeMem(fDiff);
 	    freeMem(fHtml);
 
 	    }
 	fprintf(h, "\n");
 	}
     }
 fprintf(h, "
\n"); fprintf(h, "switch to files view, user index\n"); fprintf(h, "\n\n"); fclose(h); *saveUlc = userLinesChanged; *saveUfc = userFileCount; } int slComFileCmp(const void *va, const void *vb) /* Compare two slNames. */ { const struct comFile *a = *((struct comFile **)va); const struct comFile *b = *((struct comFile **)vb); int result = strcmp(a->f->path, b->f->path); if (result == 0) result = a->commit->commitNumber - b->commit->commitNumber; return result; } void doUserFiles(char *u, struct commit *commits) /* process one user's files-view (or all if u is NULL) */ { // http://hgwdev.cse.ucsc.edu/cvs-reports/branch/user/galt/index-by-file.html // if u is NULL // http://hgwdev.cse.ucsc.edu/cvs-reports/branch/file/index.html char userPath[1024]; if (u) safef(userPath, sizeof(userPath), "%s/%s/%s/%s/index-by-file.html", outDir, outPrefix, "user", u); else safef(userPath, sizeof(userPath), "%s/%s/%s/index.html", outDir, outPrefix, "file"); FILE *h = mustOpen(userPath, "w"); if (u) { fprintf(h, "\n\nFile Changes for %s\n\n\n", u); fprintf(h, "

File Changes for %s

\n", u); fprintf(h, "switch to commits view, user index"); } else { fprintf(h, "\n\nAll File Changes\n\n\n"); fprintf(h, "

All File Changes

\n"); } fprintf(h, "

%s to %s (%s to %s) %s

\n", startTag, endTag, startDate, endDate, title); fprintf(h, "
\n");
 
 
 int totalLinesChanged = 0;
 int totalFileCount = 0;   // TODO should this count duplicates or not
 
 char *cDiff = NULL, *cHtml = NULL, *fDiff = NULL, *fHtml = NULL;
 char *relativePath = NULL;
 
 struct commit *c = NULL;
 struct files *f = NULL;
 
 struct comFile *comFiles = NULL, *cf = NULL;
 
 // pre-filter for u if u is not NULL  
 for(c = commits; c; c = c->next)
     {
     if (!u || (u && sameString(c->author, u)))
 	{
 	for(f = c->files; f; f = f->next)
 	    {
 	    AllocVar(cf);
 	    cf->f = f;
 	    cf->commit = c;
 	    slAddHead(&comFiles, cf);
 	    }
 	}
     }
 // sort by file path, and then by reverse commitNumber
 //  so that newest commit is on top.
 slSort(&comFiles, slComFileCmp);
 
 char *lastPath = "";
 
 for(cf = comFiles; cf; cf = cf->next)
     {
     c = cf->commit;
     f = cf->f;
  
     if (!sameString(f->path, lastPath))
 	{
 	lastPath = f->path;
 	fprintf(h, "%s\n", f->path);
 	}
 
     char path[1024];
 
     // context unified
     if (u)
 	safef(path, sizeof(path), "%s/%s%s", "context", f->path, c->commitId);
     else
 	safef(path, sizeof(path), "../user/%s/%s/%s%s", c->author, "context", f->path, c->commitId);
     relativePath = cloneString(path);
     safef(path, sizeof(path), "%s.html", relativePath);
     cHtml = cloneString(path);
     safef(path, sizeof(path), "%s.diff", relativePath);
     cDiff = cloneString(path);
 
 
 
     // full text (up to 10,000 lines)
     freeMem(relativePath);
     if (u)
 	safef(path, sizeof(path), "%s/%s%s", "full", f->path, c->commitId);
     else
 	safef(path, sizeof(path), "../user/%s/%s/%s%s", c->author, "context", f->path, c->commitId);
     relativePath = cloneString(path);
     safef(path, sizeof(path), "%s.html", relativePath);
     fHtml = cloneString(path);
     safef(path, sizeof(path), "%s.diff", relativePath);
     fDiff = cloneString(path);
 
     // make file view links
     fprintf(h, "  %s - lines changed %d, "
 	"context: html, text, "
 	"full: html, text\n"
 	, c->commitId, f->linesChanged
 	, cHtml, cDiff, fHtml, fDiff);
 
     //fprintf(h, "  %s\n", c->commitId);
     //fprintf(h, "  %s\n", c->date);
     fprintf(h, "    %s\n", c->comment);
 
     freeMem(relativePath);
     freeMem(cDiff);
     freeMem(cHtml);
     freeMem(fDiff);
     freeMem(fHtml);
 
     totalLinesChanged += f->linesChanged;
     ++totalFileCount;
 
     fprintf(h, "\n");
     }
 if (u)
     {
     fprintf(h, "
\n"); fprintf(h, "switch to commits view, user index"); } else { fprintf(h, "\n lines changed: %d\n files changed: %d\n", totalLinesChanged, totalFileCount); fprintf(h, "\n"); } fprintf(h, "\n\n"); fclose(h); } void doMainIndex() /* Create simple main index page */ { char path[256]; safef(path, sizeof(path), "%s/%s/index.html", outDir, outPrefix); FILE *h = mustOpen(path, "w"); fprintf(h, "\n\nSource Code Changes\n\n\n"); fprintf(h, "

%s %s Changes

\n", title, outPrefix); fprintf(h, "

%s to %s (%s to %s) %s

\n", startTag, endTag, startDate, endDate, title); fprintf(h, "
\n");
 
 fprintf(h, "  Changes by User\n");
 fprintf(h, "\n");
 fprintf(h, "  All File Changes\n");
 
 fprintf(h, "
\n\n\n"); fclose(h); } void gitReports() /* Generate code-review reports from git repo */ { int totalChangedLines = 0; int totalChangedFiles = 0; int userChangedLines = 0; int userChangedFiles = 0; tempMakeDiffName = cloneString(rTempName("/tmp", "makeDiff", ".tmp")); /* read the commits */ struct commit *commits = getCommits(), *c = NULL; /* make the user list */ for(c = commits; c; c = c->next) { if (!hashLookup(userHash, c->author)) { hashStore(userHash, c->author); struct slName *name = newSlName(c->author); slAddHead(&users, name); } } slNameSort(&users); /* create prefix dir */ char path[256]; safef(path, sizeof(path), "%s/%s", outDir, outPrefix); if (!fileExists(path) && mkdir(path, S_IRWXU | S_IRWXG | S_IROTH | S_IXOTH) != 0) errnoAbort("unable to mkdir %s", path); /* create file dir */ safef(path, sizeof(path), "%s/%s/%s", outDir, outPrefix, "file"); if (!fileExists(path) && mkdir(path, S_IRWXU | S_IRWXG | S_IROTH | S_IXOTH) != 0) errnoAbort("unable to mkdir %s", path); /* create user dir */ safef(path, sizeof(path), "%s/%s/%s", outDir, outPrefix, "user"); if (!fileExists(path) && mkdir(path, S_IRWXU | S_IRWXG | S_IROTH | S_IXOTH) != 0) errnoAbort("unable to mkdir %s", path); char usersPath[1024]; safef(usersPath, sizeof(usersPath), "%s/%s/%s/index.html", outDir, outPrefix, "user"); FILE *h = mustOpen(usersPath, "w"); fprintf(h, "\n\nChanges By User\n\n\n"); fprintf(h, "

Changes By User

\n"); fprintf(h, "

%s to %s (%s to %s) %s

\n", startTag, endTag, startDate, endDate, title); fprintf(h, "
\n");
 
 
 
 struct slName*u;
 for(u = users; u; u = u->next)
     {
     printf("user: %s\n", u->name);
 
     /* create user/name dir */
     safef(path, sizeof(path), "%s/%s/%s/%s", outDir, outPrefix, "user", u->name);
     if (!fileExists(path) && mkdir(path, S_IRWXU | S_IRWXG | S_IROTH | S_IXOTH) != 0)
 	errnoAbort("unable to mkdir %s", path);
 
     /* create user/name/context dir */
     safef(path, sizeof(path), "%s/%s/%s/%s/%s", outDir, outPrefix, "user", u->name, "context");
     if (!fileExists(path) && mkdir(path, S_IRWXU | S_IRWXG | S_IROTH | S_IXOTH) != 0)
 	errnoAbort("unable to mkdir %s", path);
 
     /* create user/name/full dir */
     safef(path, sizeof(path), "%s/%s/%s/%s/%s", outDir, outPrefix, "user", u->name, "full");
     if (!fileExists(path) && mkdir(path, S_IRWXU | S_IRWXG | S_IROTH | S_IXOTH) != 0)
 	errnoAbort("unable to mkdir %s", path);
 
     userChangedLines = 0;
     userChangedFiles = 0;
 
     // DEBUG REMOVE
     //if (sameString(u->name, "galt"))
     /* make user's reports */
     doUserCommits(u->name, commits, &userChangedLines, &userChangedFiles);
 
     doUserFiles(u->name, commits);
 
     char relPath[1024];
     safef(relPath, sizeof(relPath), "%s/index.html", u->name);
     fprintf(h, "  %s - changed lines: %d, files: %d\n", relPath, u->name, userChangedLines, userChangedFiles);
 
     totalChangedLines += userChangedLines;
     totalChangedFiles += userChangedFiles;  
     // TODO do we have to worry about counting the same file more than once?
 
     }
 
 fprintf(h, "\n  lines changed: %d\n  files changed: %d\n", totalChangedLines, totalChangedFiles);
 
 fprintf(h, "
\n\n\n"); fclose(h); // make index of all files view doUserFiles(NULL, commits); // make main index page doMainIndex(); // tidying up unlink(tempMakeDiffName); freez(&tempMakeDiffName); } int main(int argc, char *argv[]) { optionInit(&argc, argv, options); if (argc != 9) usage("wrong number of args"); if (optionExists("-help")) usage("help"); //if (optionExists("altHeader") && optionExists("autoBoundary")) // altHeader = optionVal("altHeader",altHeader); startTag = argv[1]; endTag = argv[2]; startDate = argv[3]; endDate = argv[4]; title = argv[5]; repoDir = argv[6]; outDir = argv[7]; outPrefix = argv[8]; userHash = hashNew(5); chdir(repoDir); gitReports(); hashFree(&userHash); printf("Done.\n"); return 0; }