Merge branch 'ENH_text_consumer' of git://github.com/jat255/paperless into jat255-ENH_text_consumer

src/documents/models.py

@@ -192,7 +192,11 @@ class Document(models.Model):
     TYPE_JPG = "jpg"
     TYPE_GIF = "gif"
     TYPE_TIF = "tiff"
-    TYPES = (TYPE_PDF, TYPE_PNG, TYPE_JPG, TYPE_GIF, TYPE_TIF,)
+    TYPE_TXT = "txt"
+    TYPE_CSV = "csv"
+    TYPE_MD  = "md"
+    TYPES = (TYPE_PDF, TYPE_PNG, TYPE_JPG, TYPE_GIF, TYPE_TIF,
+             TYPE_TXT, TYPE_CSV, TYPE_MD)
 
     STORAGE_TYPE_UNENCRYPTED = "unencrypted"
     STORAGE_TYPE_GPG = "gpg"
@@ -365,51 +369,52 @@ class FileInfo:
         )
     )
 
+    formats = "pdf|jpe?g|png|gif|tiff?|te?xt|md|csv"
     REGEXES = OrderedDict([
         ("created-correspondent-title-tags", re.compile(
             r"^(?P<created>\d\d\d\d\d\d\d\d(\d\d\d\d\d\d)?Z) - "
             r"(?P<correspondent>.*) - "
             r"(?P<title>.*) - "
             r"(?P<tags>[a-z0-9\-,]*)"
-            r"\.(?P<extension>pdf|jpe?g|png|gif|tiff?)$",
+            r"\.(?P<extension>{})$".format(formats),
             flags=re.IGNORECASE
         )),
         ("created-title-tags", re.compile(
             r"^(?P<created>\d\d\d\d\d\d\d\d(\d\d\d\d\d\d)?Z) - "
             r"(?P<title>.*) - "
             r"(?P<tags>[a-z0-9\-,]*)"
-            r"\.(?P<extension>pdf|jpe?g|png|gif|tiff?)$",
+            r"\.(?P<extension>{})$".format(formats),
             flags=re.IGNORECASE
         )),
         ("created-correspondent-title", re.compile(
             r"^(?P<created>\d\d\d\d\d\d\d\d(\d\d\d\d\d\d)?Z) - "
             r"(?P<correspondent>.*) - "
             r"(?P<title>.*)"
-            r"\.(?P<extension>pdf|jpe?g|png|gif|tiff?)$",
+            r"\.(?P<extension>{})$".format(formats),
             flags=re.IGNORECASE
         )),
         ("created-title", re.compile(
             r"^(?P<created>\d\d\d\d\d\d\d\d(\d\d\d\d\d\d)?Z) - "
             r"(?P<title>.*)"
-            r"\.(?P<extension>pdf|jpe?g|png|gif|tiff?)$",
+            r"\.(?P<extension>{})$".format(formats),
             flags=re.IGNORECASE
         )),
         ("correspondent-title-tags", re.compile(
             r"(?P<correspondent>.*) - "
             r"(?P<title>.*) - "
             r"(?P<tags>[a-z0-9\-,]*)"
-            r"\.(?P<extension>pdf|jpe?g|png|gif|tiff?)$",
+            r"\.(?P<extension>{})$".format(formats),
             flags=re.IGNORECASE
         )),
         ("correspondent-title", re.compile(
             r"(?P<correspondent>.*) - "
             r"(?P<title>.*)?"
-            r"\.(?P<extension>pdf|jpe?g|png|gif|tiff?)$",
+            r"\.(?P<extension>{})$".format(formats),
             flags=re.IGNORECASE
         )),
         ("title", re.compile(
             r"(?P<title>.*)"
-            r"\.(?P<extension>pdf|jpe?g|png|gif|tiff?)$",
+            r"\.(?P<extension>{})$".format(formats),
             flags=re.IGNORECASE
         ))
     ])

src/documents/parsers.py

@@ -1,9 +1,24 @@
 import logging
 import shutil
 import tempfile
+import re
 
 from django.conf import settings
 
+# This regular expression will try to find dates in the document at
+# hand and will match the following formats:
+# - XX.YY.ZZZZ with XX + YY being 1 or 2 and ZZZZ being 2 or 4 digits
+# - XX/YY/ZZZZ with XX + YY being 1 or 2 and ZZZZ being 2 or 4 digits
+# - XX-YY-ZZZZ with XX + YY being 1 or 2 and ZZZZ being 2 or 4 digits
+# - XX. MONTH ZZZZ with XX being 1 or 2 and ZZZZ being 2 or 4 digits
+# - MONTH ZZZZ, with ZZZZ being 4 digits
+# - MONTH XX, ZZZZ with XX being 1 or 2 and ZZZZ being 4 digits
+pattern = re.compile(
+    r'\b([0-9]{1,2})[\.\/-]([0-9]{1,2})[\.\/-]([0-9]{4}|[0-9]{2})\b|' +
+    r'\b([0-9]{1,2}[\. ]+[^ ]{3,9} ([0-9]{4}|[0-9]{2}))\b|' +
+    r'\b([^\W\d_]{3,9} [0-9]{1,2}, ([0-9]{4}))\b|' +
+    r'\b([^\W\d_]{3,9} [0-9]{4})\b')
+
 
 class ParseError(Exception):
     pass

src/documents/views.py

@@ -48,6 +48,9 @@ class FetchView(SessionOrBasicAuthMixin, DetailView):
             Document.TYPE_JPG: "image/jpeg",
             Document.TYPE_GIF: "image/gif",
             Document.TYPE_TIF: "image/tiff",
+            Document.TYPE_CSV: "text/csv",
+            Document.TYPE_MD:  "text/markdown",
+            Document.TYPE_TXT: "text/plain"
         }
 
         if self.kwargs["kind"] == "thumb":

src/paperless/settings.py

@@ -67,6 +67,7 @@ INSTALLED_APPS = [
     "documents.apps.DocumentsConfig",
     "reminders.apps.RemindersConfig",
     "paperless_tesseract.apps.PaperlessTesseractConfig",
+    "paperless_text.apps.PaperlessTextConfig",
 
     "django.contrib.admin",
 

src/paperless_tesseract/parsers.py

@@ -14,7 +14,7 @@ from pyocr.libtesseract.tesseract_raw import \
 from pyocr.tesseract import TesseractError
 
 import pdftotext
-from documents.parsers import DocumentParser, ParseError
+from documents.parsers import DocumentParser, ParseError, pattern
 
 from .languages import ISO639
 
@@ -210,20 +210,6 @@ class RasterisedDocumentParser(DocumentParser):
         except ParseError as e:
             return None
 
-        # This regular expression will try to find dates in the document at
-        # hand and will match the following formats:
-        # - XX.YY.ZZZZ with XX + YY being 1 or 2 and ZZZZ being 2 or 4 digits
-        # - XX/YY/ZZZZ with XX + YY being 1 or 2 and ZZZZ being 2 or 4 digits
-        # - XX-YY-ZZZZ with XX + YY being 1 or 2 and ZZZZ being 2 or 4 digits
-        # - XX. MONTH ZZZZ with XX being 1 or 2 and ZZZZ being 2 or 4 digits
-        # - MONTH ZZZZ, with ZZZZ being 4 digits
-        # - MONTH XX, ZZZZ with XX being 1 or 2 and ZZZZ being 4 digits
-        pattern = re.compile(
-            r'\b([0-9]{1,2})[\.\/-]([0-9]{1,2})[\.\/-]([0-9]{4}|[0-9]{2})\b|' +
-            r'\b([0-9]{1,2}[\. ]+[^ ]{3,9} ([0-9]{4}|[0-9]{2}))\b|' +
-            r'\b([^\W\d_]{3,9} [0-9]{1,2}, ([0-9]{4}))\b|' +
-            r'\b([^\W\d_]{3,9} [0-9]{4})\b')
-
         # Iterate through all regex matches and try to parse the date
         for m in re.finditer(pattern, text):
             datestring = m.group(0)

src/paperless_text/apps.py

@@ -0,0 +1,16 @@
+from django.apps import AppConfig
+
+
+class PaperlessTextConfig(AppConfig):
+
+    name = "paperless_text"
+
+    def ready(self):
+
+        from documents.signals import document_consumer_declaration
+
+        from .signals import ConsumerDeclaration
+
+        document_consumer_declaration.connect(ConsumerDeclaration.handle)
+
+        AppConfig.ready(self)

src/paperless_text/parsers.py

@@ -0,0 +1,131 @@
+import os
+import re
+import subprocess
+
+import dateparser
+from django.conf import settings
+
+from documents.parsers import DocumentParser, ParseError, pattern
+
+
+class TextDocumentParser(DocumentParser):
+    """
+    This parser directly parses a text document (.txt, .md, or .csv)
+    """
+
+    CONVERT = settings.CONVERT_BINARY
+    THREADS = int(settings.OCR_THREADS) if settings.OCR_THREADS else None
+    UNPAPER = settings.UNPAPER_BINARY
+    DATE_ORDER = settings.DATE_ORDER
+    DEFAULT_OCR_LANGUAGE = settings.OCR_LANGUAGE
+    OCR_ALWAYS = settings.OCR_ALWAYS
+
+    def __init__(self, path):
+        super().__init__(path)
+        self._text = None
+
+    def get_thumbnail(self):
+        """
+        The thumbnail of a txt is just a 500px wide image of the text
+        rendered onto a letter-sized page.
+        """
+        # The below is heavily cribbed from https://askubuntu.com/a/590951
+
+        bg_color = "white"  # bg color
+        text_color = "black"  # text color
+        psize = [500, 647]  # icon size
+        n_lines = 50  # number of lines to show
+        output_file = os.path.join(self.tempdir, "convert-txt.png")
+
+        temp_bg = os.path.join(self.tempdir, "bg.png")
+        temp_txlayer = os.path.join(self.tempdir, "tx.png")
+        picsize = "x".join([str(n) for n in psize])
+        txsize = "x".join([str(n - 8) for n in psize])
+
+        def create_bg():
+            work_size = ",".join([str(n - 1) for n in psize])
+            r = str(round(psize[0] / 10));
+            rounded = ",".join([r, r])
+            run_command(self.CONVERT, "-size ", picsize, ' xc:none -draw ',
+                        '"fill ', bg_color, ' roundrectangle 0,0,',
+                        work_size, ",", rounded, '" ', temp_bg)
+
+        def read_text():
+            with open(self.document_path, 'r') as src:
+                lines = [l.strip() for l in src.readlines()]
+                text = "\n".join([l for l in lines[:n_lines]])
+                return text.replace('"', "'")
+
+        def create_txlayer():
+            run_command(self.CONVERT,
+                        "-background none",
+                        "-fill",
+                        text_color,
+                        "-pointsize", "12",
+                        "-border 4 -bordercolor none",
+                        "-size ", txsize,
+                        ' caption:"', read_text(), '" ',
+                        temp_txlayer)
+
+        create_txlayer()
+        create_bg()
+        run_command(self.CONVERT, temp_bg, temp_txlayer,
+                    "-background None -layers merge ", output_file)
+
+        return output_file
+
+    def get_text(self):
+
+        if self._text is not None:
+            return self._text
+
+        with open(self.document_path, 'r') as f:
+            self._text = f.read()
+
+        return self._text
+
+    def get_date(self):
+        date = None
+        datestring = None
+
+        try:
+            text = self.get_text()
+        except ParseError as e:
+            return None
+
+        # Iterate through all regex matches and try to parse the date
+        for m in re.finditer(pattern, text):
+            datestring = m.group(0)
+
+            try:
+                date = dateparser.parse(
+                           datestring,
+                           settings={'DATE_ORDER': self.DATE_ORDER,
+                                     'PREFER_DAY_OF_MONTH': 'first',
+                                     'RETURN_AS_TIMEZONE_AWARE': True})
+            except TypeError:
+                # Skip all matches that do not parse to a proper date
+                continue
+
+            if date is not None:
+                break
+
+        if date is not None:
+            self.log("info", "Detected document date " + date.isoformat() +
+                             " based on string " + datestring)
+        else:
+            self.log("info", "Unable to detect date for document")
+
+        return date
+
+
+def run_command(*args):
+    environment = os.environ.copy()
+    if settings.CONVERT_MEMORY_LIMIT:
+        environment["MAGICK_MEMORY_LIMIT"] = settings.CONVERT_MEMORY_LIMIT
+    if settings.CONVERT_TMPDIR:
+        environment["MAGICK_TMPDIR"] = settings.CONVERT_TMPDIR
+
+    if not subprocess.Popen(' '.join(args), env=environment,
+                            shell=True).wait() == 0:
+        raise ParseError("Convert failed at {}".format(args))

src/paperless_text/signals.py

@@ -0,0 +1,23 @@
+import re
+
+from .parsers import TextDocumentParser
+
+
+class ConsumerDeclaration:
+
+    MATCHING_FILES = re.compile("^.*\.(te?xt|md|csv)$")
+
+    @classmethod
+    def handle(cls, sender, **kwargs):
+        return cls.test
+
+    @classmethod
+    def test(cls, doc):
+
+        if cls.MATCHING_FILES.match(doc.lower()):
+            return {
+                "parser": TextDocumentParser,
+                "weight": 10
+            }
+
+        return None