Feat: add image preview to retrieval test. (#7610)

### What problem does this PR solve?

#7608

### Type of change

- [x] New Feature (non-breaking change which adds functionality)

rag/app/manual.py

@@ -265,7 +265,9 @@ def chunk(filename, binary=None, from_page=0, to_page=100000,
         res = tokenize_table(tbls, doc, eng)
         for text, image in ti_list:
             d = copy.deepcopy(doc)
-            d['image'] = image
+            if image:
+                d['image'] = image
+                d["doc_type_kwd"] = "image"
             tokenize(d, text, eng)
             res.append(d)
         return res

rag/app/picture.py

@@ -36,7 +36,8 @@ def chunk(filename, binary, tenant_id, lang, callback=None, **kwargs):
     doc = {
         "docnm_kwd": filename,
         "title_tks": rag_tokenizer.tokenize(re.sub(r"\.[a-zA-Z]+$", "", filename)),
-        "image": img
+        "image": img,
+        "doc_type_kwd": "image"
     }
     bxs = ocr(np.array(img))
     txt = "\n".join([t[0] for _, t in bxs if t[0]])

rag/app/presentation.py

@@ -115,6 +115,7 @@ def chunk(filename, binary=None, from_page=0, to_page=100000,
             d = copy.deepcopy(doc)
             pn += from_page
             d["image"] = img
+            d["doc_type_kwd"] = "image"
             d["page_num_int"] = [pn + 1]
             d["top_int"] = [0]
             d["position_int"] = [(pn + 1, 0, img.size[0], 0, img.size[1])]

rag/app/qa.py

@@ -271,7 +271,9 @@ def beAdocPdf(d, q, a, eng, image, poss):
         [qprefix + rmPrefix(q), aprefix + rmPrefix(a)])
     d["content_ltks"] = rag_tokenizer.tokenize(q)
     d["content_sm_ltks"] = rag_tokenizer.fine_grained_tokenize(d["content_ltks"])
-    d["image"] = image
+    if image:
+        d["image"] = image
+        d["doc_type_kwd"] = "image"
     add_positions(d, poss)
     return d
 
@@ -283,7 +285,9 @@ def beAdocDocx(d, q, a, eng, image, row_num=-1):
         [qprefix + rmPrefix(q), aprefix + rmPrefix(a)])
     d["content_ltks"] = rag_tokenizer.tokenize(q)
     d["content_sm_ltks"] = rag_tokenizer.fine_grained_tokenize(d["content_ltks"])
-    d["image"] = image
+    if image:
+        d["image"] = image
+        d["doc_type_kwd"] = "image"
     if row_num >= 0:
         d["top_int"] = [row_num]
     return d