From 012249ceca8a8039dc5d093b3a5679b48039489f Mon Sep 17 00:00:00 2001
From: Dima Gerasimov <karlicoss@gmail.com>
Date: Wed, 8 Jan 2020 22:07:34 +0000
Subject: [PATCH] Use experimental cachew Exceptions

---
 my/common.py |  2 ++
 my/pdfs.py   | 19 ++++++++++++++-----
 2 files changed, 16 insertions(+), 5 deletions(-)

diff --git a/my/common.py b/my/common.py
index e1389e3..861e48b 100644
--- a/my/common.py
+++ b/my/common.py
@@ -125,4 +125,6 @@ def mcachew(*args, **kwargs):
         warnings.warn('cachew library not found. You might want to install it to speed things up. See https://github.com/karlicoss/cachew')
         return lambda orig_func: orig_func
     else:
+        import cachew.experimental
+        cachew.experimental.enable_exceptions()  # TODO do it only once?
         return cachew.cachew(*args, **kwargs)
diff --git a/my/pdfs.py b/my/pdfs.py
index 2191f7b..67d7aa0 100755
--- a/my/pdfs.py
+++ b/my/pdfs.py
@@ -23,6 +23,7 @@ def get_logger():
 
 
 def is_ignored(p: Path) -> bool:
+    # ignore some extremely heavy files
     return paths.pdfs.is_ignored(p)
 
 
@@ -85,19 +86,22 @@ def get_annots(p: Path) -> List[Annotation]:
     with p.open('rb') as fo:
         f = io.StringIO()
         with redirect_stderr(f):
-            # TODO FIXME defensive, try on garbage file (s)
             (annots, outlines) = pdfannots.process_file(fo, emit_progress=False)
             # outlines are kinda like TOC, I don't really need them
     return [as_annotation(raw_ann=a, path=str(p)) for a in annots]
     # TODO stderr?
 
 
-# TODO cachew needs to be based on mtime, hence take candidates, not roots
-# @mcachew
-def iter_annotations(roots=None) -> Iterator[Res[Annotation]]:
+def hash_files(pdfs: List[Path]):
+    # if mtime hasn't changed then the file hasn't changed either
+    return [(pdf, pdf.stat().st_mtime) for pdf in pdfs]
+
+# TODO might make more sense to be more fine grained here, e.g. cache annotations for indifidual files
+
+@mcachew(hashf=hash_files)
+def _iter_annotations(pdfs: List[Path]) -> Iterator[Res[Annotation]]:
     logger = get_logger()
 
-    pdfs = list(sorted(candidates(roots=roots)))
     logger.info('processing %d pdfs', len(pdfs))
 
     # TODO how to print to stdout synchronously?
@@ -117,6 +121,11 @@ def iter_annotations(roots=None) -> Iterator[Res[Annotation]]:
                 yield e
 
 
+def iter_annotations(roots=None) -> Iterator[Res[Annotation]]:
+    pdfs = list(sorted(candidates(roots=roots)))
+    yield from _iter_annotations(pdfs=pdfs)
+
+
 class Pdf(NamedTuple):
     path: Path
     annotations: List[Annotation]