# HG changeset patch
# User Valentin Gatien-Baron <valentin.gatienbaron@gmail.com>
# Date 1589754825 14400
# Node ID 708ad5cf5e5a7666ec6619829a68c6ef8d4cb3e0
# Parent  8d552701806d537bffe741b6ecebc4bbe7226329
grep: grep the working copy faster

`hg grep qqqq` in the mercurial repo:
before: 0,859s
after: 0,233s

`hg grep somethingwithnomatch` in mozilla-central:
before: 51s
after: 19s

This is probably also a tiny bug fix, because the code was looking up
a node for filename `pfn` on a filelog for filename `fn`, which are
most of the time the same filename, but don't have to be.

Ignoring performance and the bug fix, the code should have the same
behavior.

Differential Revision: https://phab.mercurial-scm.org/D8545

diff -r 8d552701806d -r 708ad5cf5e5a mercurial/commands.py
--- a/mercurial/commands.py	Sun May 17 13:10:54 2020 -0400
+++ b/mercurial/commands.py	Sun May 17 18:33:45 2020 -0400
@@ -3578,18 +3578,27 @@
 
     getrenamed = scmutil.getrenamedfn(repo)
 
-    def get_file_content(filename, filelog, filenode, context, revision):
-        try:
-            content = filelog.read(filenode)
-        except error.WdirUnsupported:
-            content = context[filename].data()
-        except error.CensoredNodeError:
-            content = None
-            ui.warn(
-                _(b'cannot search in censored file: %(filename)s:%(revnum)s\n')
-                % {b'filename': filename, b'revnum': pycompat.bytestr(revision)}
-            )
-        return content
+    def readfile(ctx, fn):
+        rev = ctx.rev()
+        if rev is None:
+            fctx = ctx[fn]
+            try:
+                return fctx.data()
+            except IOError as e:
+                if e.errno != errno.ENOENT:
+                    raise
+        else:
+            flog = getfile(fn)
+            fnode = ctx.filenode(fn)
+            try:
+                return flog.read(fnode)
+            except error.CensoredNodeError:
+                ui.warn(
+                    _(
+                        b'cannot search in censored file: %(filename)s:%(revnum)s\n'
+                    )
+                    % {b'filename': fn, b'revnum': pycompat.bytestr(rev),}
+                )
 
     def prep(ctx, fns):
         rev = ctx.rev()
@@ -3600,10 +3609,10 @@
             matches.setdefault(parent, {})
         files = revfiles.setdefault(rev, [])
         for fn in fns:
-            flog = getfile(fn)
-            try:
-                fnode = ctx.filenode(fn)
-            except error.LookupError:
+            # fn might not exist in the revision (could be a file removed by the
+            # revision). We could check `fn not in ctx` even when rev is None,
+            # but it's less racy to protect againt that in readfile.
+            if rev is not None and fn not in ctx:
                 continue
 
             copy = None
@@ -3618,20 +3627,12 @@
             files.append(fn)
 
             if fn not in matches[rev]:
-                content = get_file_content(fn, flog, fnode, ctx, rev)
-                grepbody(fn, rev, content)
+                grepbody(fn, rev, readfile(ctx, fn))
 
             if diff:
                 pfn = copy or fn
-                if pfn not in matches[parent]:
-                    try:
-                        pfnode = pctx.filenode(pfn)
-                        pcontent = get_file_content(
-                            pfn, flog, pfnode, pctx, parent
-                        )
-                        grepbody(pfn, parent, pcontent)
-                    except error.LookupError:
-                        pass
+                if pfn not in matches[parent] and pfn in pctx:
+                    grepbody(pfn, parent, readfile(pctx, pfn))
 
     ui.pager(b'grep')
     fm = ui.formatter(b'grep', opts)