From 0113fe9dd6bcaf7906621a6d214ce8f55653103d Mon Sep 17 00:00:00 2001
From: duckduckgrayduck <102841251+duckduckgrayduck@users.noreply.github.com>
Date: Mon, 12 Aug 2024 13:27:09 -0500
Subject: [PATCH] Add page number identification for legal citations

---
 main.py          | 27 ++++++++++++++++-----------
 requirements.txt |  2 +-
 2 files changed, 17 insertions(+), 12 deletions(-)

diff --git a/main.py b/main.py
index 8d2696f..22bb409 100644
--- a/main.py
+++ b/main.py
@@ -1,24 +1,29 @@
+""" Requires eyecite to find legal citations """
+import csv
 from documentcloud.addon import AddOn
 from eyecite import get_citations
-import csv
+
 
 class LegalCitations(AddOn):
+    """ DocumentCloud Add-On that uses eyecite to find legal citations in a document """
     def main(self):
-
+        """ Loops through each page on each document """
         citations_found = []
-     
+
         for document in self.get_documents():
-            citation_list = get_citations(document.full_text)
-            tagged_citation_list = [
-                (document.title, document.id, citation)
-                for citation in citation_list
-            ]
-            citations_found += tagged_citation_list
+            for page_number in range(1, document.page_count + 1):
+                page_text = document.get_page_text(page_number)
+                citation_list = get_citations(page_text)
+                tagged_citation_list = [
+                    (document.title, document.id, page_number, citation)
+                    for citation in citation_list
+                ]
+                citations_found += tagged_citation_list
 
         # output the citations as a CSV.
-        with open("citations_found.csv", "w+") as file_:
+        with open("citations_found.csv", "w+", encoding="utf-8") as file_:
             writer = csv.writer(file_)
-            writer.writerow(("title", "id", "citation"))
+            writer.writerow(("title", "id", "page number", "citation"))
             writer.writerows(citations_found)
             self.upload_file(file_)
 
diff --git a/requirements.txt b/requirements.txt
index b6c7255..0b010ba 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,3 +1,3 @@
-python-documentcloud==3.7.1
+python-documentcloud==4.1.3
 requests
 eyecite