AI/ragflow
1
0
Fork 0
mirror of https://git.mirrors.martin98.com/https://github.com/infiniflow/ragflow.git synced 2025-07-31 08:42:00 +08:00
Code Issues Packages Projects Releases Wiki Activity

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

Browse Source 
### What problem does this PR solve?

#7608

### Type of change

- [x] New Feature (non-breaking change which adds functionality)
This commit is contained in:
Kevin Hu 2025-05-13 14:30:36 +08:00 committed by GitHub
parent 5c9025918a
commit 321a280031
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
7 changed files with 21 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
conf/infinity_mapping.json
View File

@ -38,5 +38,7 @@
"entity_type_kwd": {"type": "varchar", "default": "", "analyzer": "whitespace-#"},
"source_id": {"type": "varchar", "default": "", "analyzer": "whitespace-#"},
"n_hop_with_weight": {"type": "varchar", "default": ""},
"removed_kwd": {"type": "varchar", "default": "", "analyzer": "whitespace-#"}
"removed_kwd": {"type": "varchar", "default": "", "analyzer": "whitespace-#"},
"doc_type_kwd": {"type": "varchar", "default": "", "analyzer": "whitespace-#"}
}

4
rag/app/manual.py
View File

@ -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

3
rag/app/picture.py
View File

@ -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]])

1
rag/app/presentation.py
View File

@ -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])]

8
rag/app/qa.py
View File

@ -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

5
rag/nlp/__init__.py
View File

@ -301,6 +301,7 @@ def tokenize_table(tbls, doc, eng, batch_size=10):
d["content_with_weight"] = rows
if img:
d["image"] = img
d["doc_type_kwd"] = "image"
if poss:
add_positions(d, poss)
res.append(d)
@ -310,7 +311,9 @@ def tokenize_table(tbls, doc, eng, batch_size=10):
d = copy.deepcopy(doc)
r = de.join(rows[i:i + batch_size])
tokenize(d, r, eng)
d["image"] = img
if img:
d["image"] = img
d["doc_type_kwd"] = "image"
add_positions(d, poss)
res.append(d)
return res

3
rag/nlp/search.py
View File

@ -83,7 +83,7 @@ class Dealer:
src = req.get("fields",
["docnm_kwd", "content_ltks", "kb_id", "img_id", "title_tks", "important_kwd", "position_int",
"doc_id", "page_num_int", "top_int", "create_timestamp_flt", "knowledge_graph_kwd",
"question_kwd", "question_tks",
"question_kwd", "question_tks", "doc_type_kwd",
"available_int", "content_with_weight", PAGERANK_FLD, TAG_FLD])
kwds = set([])
@ -417,6 +417,7 @@ class Dealer:
"term_similarity": tsim[i],
"vector": chunk.get(vector_column, zero_vector),
"positions": position_int,
"doc_type_kwd": chunk.get("doc_type_kwd", "")
}
if highlight and sres.highlight:
if id in sres.highlight: