Merge branch 'master' of github.com:SimonEff/Know-how-Management-OLL

SimonEff · Mar 25, 2024 · d945cc4 · d945cc4
2 parents 7a77be8 + eddd20e
commit d945cc4
Show file tree

Hide file tree

Showing 4 changed files with 42 additions and 20 deletions.
diff --git a/autotagger/create_tags.py b/autotagger/create_tags.py
@@ -10,18 +10,25 @@
 
 def iter_texts(data_dir=DATA_DIR):
     for path in data_dir.glob("**/*.*"):
+        parts = []
         try:
             if path.suffix == '.docx':
                 doc = docx.Document(str(path))
-                text = '\n\n'.join(para.text for para in doc.paragraphs)
-                yield path, text
+                for i, para in enumerate(doc.paragraphs):
+                    part = para.text
+                    parts.append(part)
+                    yield path, f"Abschnitt {i+1}, {path.name}", part
             elif path.suffix == '.pdf':
                 with path.open("rb") as fp:
                     pdf = PyPDF2.PdfReader(fp)
-                    text = '\n\n'.join(page.extract_text() for page in pdf.pages)
-                    yield path, text
+                    for i, page in enumerate(pdf.pages):
+                        part = page.extract_text()
+                        parts.append(part)
+                        yield path, f"Seite {i+1}, {path.name}", part
         except:
             pass
+        if parts:
+            yield path, path.name, '\n\n'.join(parts)
 
 def create_tags(text):
     nlp_doc = NLP(text)
@@ -31,5 +38,5 @@ def create_tags(text):
     return {'text': text, 'tags': tags}
 
 print("Create tags...")
-data = {path.name: create_tags(text) for path, text in tqdm.tqdm(iter_texts(), unit=" documents")}
+data = {name: create_tags(text) for path, name, text in tqdm.tqdm(iter_texts(), unit=" documents")}
 pathlib.Path("data.json").write_text(json.dumps(data), encoding='utf-8')
diff --git a/autotagger/searching.py b/autotagger/searching.py
@@ -16,22 +16,22 @@ def clean_tags(tags):
 
 def find_top_documents(query, maxn=MAXN, minscore=MINSCORE):
     documents = list(DATA.items())
-    scorer = BM25Okapi([clean_tags(doc['tags']) for filename, doc in documents])
+    scorer = BM25Okapi([clean_tags(doc['tags']) for name, doc in documents])
     scores = scorer.get_scores(query)
     top_docs = numpy.argsort(scores)[::-1][:MAXN]
     for i in top_docs:
-        filename, doc = documents[i]
+        name, doc = documents[i]
         score = scores[i]
         if score < MINSCORE:
             break
         yield {
-            'filename': filename,
-            'excerpt': textwrap.shorten(doc['text'], 100),
+            'name': name,
+            'excerpt': textwrap.shorten(doc['text'], 200),
             'tags': doc['tags'],
             'score': int(100*score),
         }
 
 if __name__ == '__main__':
     query = clean_tags(sys.argv[1:])
     for match in find_top_documents(query):
-        print(f"{match['filename']} (Score: {match['score']})")
+        print(f"{match['name']} (Score: {match['score']})")
diff --git a/autotagger/static/style.css b/autotagger/static/style.css
@@ -1,5 +1,5 @@
 body{
-    background-color: rgb(255, 255, 200);
+    background-color: #C4DFE6;
 }
 h1{
     text-align: center;
@@ -8,6 +8,9 @@ form{
     margin-bottom: 30px;
     text-align:center;
 }
+#query{
+    width: 40%;
+}
 .container{
     margin-left: auto;
     margin-right: auto;
@@ -23,19 +26,31 @@ form{
 }
 .card{
     display: block;
-    box-shadow: 0px 0px 15px #ccc;
+    border-radius: 10px;
+    background-image: linear-gradient(to right bottom, #86b4cd, #96c5dd, #86b4cd);
+    box-shadow: 2px 2px 7px #ccc;
     margin-left: auto;
     margin-right: auto;
     padding: 15px;
     margin-bottom: 25px;
 }
-.card a{
-    text-decoration: none;
-    padding: 2px;
-    background-color: teal;
-    color: white;
-    font-size: 36px;
+.card b{
+    font-family:Georgia, 'Times New Roman', Times, serif;
+    color: #162d3d;
+    text-overflow: ellipsis;
+    overflow: hidden;
+    white-space: nowrap;
+    max-width: 90%;
+    margin-bottom: 10px;
+    display: inline-block;
+}
+.card i{
+    font-family:Impact, Haettenschweiler, 'Arial Narrow Bold', sans-serif;
+    color: #466c8d;
 }
 .card small{
     margin-top: auto;
+    margin-left: 15px;
+    font-family:'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
+    color: #36617d;
 }
diff --git a/autotagger/templates/index.html b/autotagger/templates/index.html
@@ -9,8 +9,8 @@
     {% for result in results %}
         <div class="card">
         <div>
-                <b>{{result.filename}}</b>
-                <i class="score">Score {{result.score}}</i>
+            <i class="score">Score {{result.score}}</i>
+            <b>{{result.name}}</b>
         </div>
         <small>{{result.excerpt}}</small>
     </div>