Add formatting style Markdown

papis_extract/__init__.py

@@ -8,7 +8,7 @@ import papis.strings
 from papis.document import Document
 
 from papis_extract import extractor, exporter
-from papis_extract.annotation_data import AnnotatedDocument
+from papis_extract.annotation_data import AnnotatedDocument, Markdown
 
 logger = papis.logging.get_logger(__name__)
 
@@ -84,6 +84,6 @@ def run(
     doc_annotations: list[AnnotatedDocument] = extractor.start(documents)
 
     if write:
-        exporter.to_notes(doc_annotations, edit=edit, git=git)
+        exporter.to_notes(doc_annotations, Markdown(), edit=edit, git=git)
     else:
-        exporter.to_stdout(doc_annotations)
+        exporter.to_stdout(doc_annotations, Markdown())

papis_extract/annotation_data.py

@@ -1,5 +1,6 @@
 import math
 from dataclasses import dataclass, field
+from typing import Protocol
 
 import papis.config
 from papis.document import Document
@@ -55,9 +56,7 @@ class Annotation:
         Finds the closest named color to the annotation and returns it,
         using euclidian distance between the two color vectors.
         """
-        annot_colors = (
+        annot_colors = self.colors or (0.0, 0.0, 0.0)
-            self.colors or (0.0, 0.0, 0.0)
-        )
         nearest = None
         minimum_similarity = (
             papis.config.getfloat("minimum_similarity_color", "plugins.extract") or 1.0
@@ -85,3 +84,17 @@ class Annotation:
 class AnnotatedDocument:
     document: Document
     annotations: list[Annotation]
+
+
+@dataclass
+class Templating(Protocol):
+    string: str
+
+
+@dataclass
+class Markdown:
+    string: str = (
+        "{{#tag}}#{{tag}}\n{{/tag}}"
+        "{{#quote}}> {{quote}}{{/quote}} {{#page}}[p. {{page}}]{{/page}}\n"
+        "{{#note}}  NOTE: {{note}}{{/note}}"
+    )

papis_extract/exporter.py

@@ -7,17 +7,12 @@ import papis.git
 import papis.config
 import Levenshtein
 
-from papis_extract.annotation_data import AnnotatedDocument, Annotation
+from papis_extract.annotation_data import AnnotatedDocument, Annotation, Templating
 
 logger = papis.logging.get_logger(__name__)
 
 
-def _format_annotation(annotation: Annotation) -> str:
+def to_stdout(annots: list[AnnotatedDocument], template: Templating) -> None:
-    note = f"NOTE: {annotation.content}" if annotation.content else ""
-    return f"> {annotation.text}\n  {note}"
-
-
-def to_stdout(annots: list[AnnotatedDocument]) -> None:
     """Pretty print annotations to stdout.
 
     Gives a nice human-readable representations of
@@ -27,6 +22,7 @@ def to_stdout(annots: list[AnnotatedDocument]) -> None:
     if not annots:
         return
 
+    last = annots[-1]
     for entry in annots:
         if not entry.annotations:
             continue
@@ -39,13 +35,15 @@ def to_stdout(annots: list[AnnotatedDocument]) -> None:
             f"{title_decoration}\n{papis.document.describe(entry.document)}\n{title_decoration}\n"
         )
         for a in entry.annotations:
-            print(_format_annotation(a))
+            print(a.format(template.string))
 
-        if entry != annots[-1]:
+        if entry != last:
             print("\n")
 
 
-def to_notes(annots: list[AnnotatedDocument], edit: bool, git: bool) -> None:
+def to_notes(
+    annots: list[AnnotatedDocument], template: Templating, edit: bool, git: bool
+) -> None:
     """Write annotations into document notes.
 
     Permanently writes the given annotations into notes
@@ -61,7 +59,7 @@ def to_notes(annots: list[AnnotatedDocument], edit: bool, git: bool) -> None:
 
         formatted_annotations: list[str] = []
         for a in entry.annotations:
-            formatted_annotations.append(_format_annotation(a))
+            formatted_annotations.append(a.format(template.string))
 
         _add_annots_to_note(entry.document, formatted_annotations)
 

tests/test_formatting.py

@@ -0,0 +1,14 @@
+import chevron
+
+from papis_extract.annotation_data import Markdown
+
+def test_markdown_default():
+    fmt = Markdown()
+    assert chevron.render(fmt.string, {
+        "file": "somefile/somewhere.pdf",
+        "quote": "I am quote",
+        "note": "and including note.",
+        "page": 46,
+        "tag": "important",
+        "type": "highlight",
+    }) == "#important\n> I am quote [p. 46]\n  NOTE: and including note."