Rename page_num_list, top_list, position_list (#3940)

### What problem does this PR solve?

Rename page_num_list, top_list, position_list to page_num_int, top_int,
position_int

### Type of change

- [x] Refactoring

Dockerfile

@@ -34,9 +34,15 @@ RUN --mount=type=bind,from=infiniflow/ragflow_deps:latest,source=/,target=/deps
     cp /deps/cl100k_base.tiktoken /ragflow/9b5ad71b2ce5302211f9c61530b329a4922fc6a4
 
 ENV TIKA_SERVER_JAR="file:///ragflow/tika-server-standard-3.0.0.jar"
+ENV DEBIAN_FRONTEND=noninteractive
 
 # Setup apt
-# cv2 requires libGL.so.1
+# Python package and implicit dependencies:
+# opencv-python: libglib2.0-0 libglx-mesa0 libgl1
+# aspose-slides: pkg-config libicu-dev libgdiplus         libssl1.1_1.1.1f-1ubuntu2_amd64.deb
+# python-pptx:   default-jdk                              tika-server-standard-3.0.0.jar
+# selenium:      libatk-bridge2.0-0                       chrome-linux64-121-0-6167-85
+# Building C extensions: libpython3-dev libgtk-4-1 libnss3 xdg-utils libgbm-dev
 RUN --mount=type=cache,id=ragflow_apt,target=/var/cache/apt,sharing=locked \
     if [ "$NEED_MIRROR" == "1" ]; then \
         sed -i 's|http://archive.ubuntu.com|https://mirrors.tuna.tsinghua.edu.cn|g' /etc/apt/sources.list; \
@@ -47,8 +53,12 @@ RUN --mount=type=cache,id=ragflow_apt,target=/var/cache/apt,sharing=locked \
     apt update && \
     apt --no-install-recommends install -y ca-certificates && \
     apt update && \
-    DEBIAN_FRONTEND=noninteractive apt install -y curl libpython3-dev nginx libglib2.0-0 libglx-mesa0 pkg-config libicu-dev libgdiplus default-jdk python3-pip pipx \
-    libatk-bridge2.0-0 libgtk-4-1 libnss3 xdg-utils unzip libgbm-dev wget git nginx libgl1 vim less
+    apt install -y libglib2.0-0 libglx-mesa0 libgl1 && \
+    apt install -y pkg-config libicu-dev libgdiplus && \
+    apt install -y default-jdk && \
+    apt install -y libatk-bridge2.0-0 && \
+    apt install -y libpython3-dev libgtk-4-1 libnss3 xdg-utils libgbm-dev && \
+    apt install -y python3-pip pipx nginx unzip curl wget git vim less
 
 RUN if [ "$NEED_MIRROR" == "1" ]; then \
         pip3 config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple && \

api/apps/chunk_app.py

@@ -71,7 +71,7 @@ def list_chunk():
                 "question_kwd": sres.field[id].get("question_kwd", []),
                 "image_id": sres.field[id].get("img_id", ""),
                 "available_int": int(sres.field[id].get("available_int", 1)),
-                "positions": json.loads(sres.field[id].get("position_list", "[]")),
+                "positions": sres.field[id].get("position_int", []),
             }
             assert isinstance(d["positions"], list)
             assert len(d["positions"]) == 0 or (isinstance(d["positions"][0], list) and len(d["positions"][0]) == 5)

api/apps/sdk/doc.py

@@ -846,7 +846,7 @@ def list_chunks(tenant_id, dataset_id, document_id):
                 "question_kwd": sres.field[id].get("question_kwd", []),
                 "img_id": sres.field[id].get("img_id", ""),
                 "available_int": sres.field[id].get("available_int", 1),
-                "positions": sres.field[id].get("position_int", "").split("\t"),
+                "positions": sres.field[id].get("position_int", []),
             }
             if len(d["positions"]) % 5 == 0:
                 poss = []

conf/infinity_mapping.json

@@ -16,9 +16,9 @@
 	"content_with_weight": {"type": "varchar", "default": ""},
 	"content_ltks": {"type": "varchar", "default": ""},
 	"content_sm_ltks": {"type": "varchar", "default": ""},
-	"page_num_list": {"type": "varchar", "default": ""},
-	"top_list": {"type": "varchar", "default": ""},
-	"position_list": {"type": "varchar", "default": ""},
+	"page_num_int": {"type": "varchar", "default": ""},
+	"top_int": {"type": "varchar", "default": ""},
+	"position_int": {"type": "varchar", "default": ""},
 	"weight_int": {"type": "integer", "default": 0},
 	"weight_flt": {"type": "float", "default": 0.0},
 	"rank_int": {"type": "integer", "default": 0},

graphrag/search.py

@@ -58,7 +58,7 @@ class KGSearch(Dealer):
         matchDense = self.get_vector(qst, emb_mdl, 1024, req.get("similarity", 0.1))
         q_vec = matchDense.embedding_data
         src = req.get("fields", ["docnm_kwd", "content_ltks", "kb_id", "img_id", "title_tks", "important_kwd",
-                                 "doc_id", f"q_{len(q_vec)}_vec", "position_list", "name_kwd",
+                                 "doc_id", f"q_{len(q_vec)}_vec", "position_int", "name_kwd",
                                  "available_int", "content_with_weight",
                                  "weight_int", "weight_flt"
                                  ])

rag/app/presentation.py

@@ -20,7 +20,6 @@ from rag.nlp import tokenize, is_english
 from rag.nlp import rag_tokenizer
 from deepdoc.parser import PdfParser, PptParser, PlainParser
 from PyPDF2 import PdfReader as pdf2_read
-import json
 
 
 class Ppt(PptParser):
@@ -109,9 +108,9 @@ def chunk(filename, binary=None, from_page=0, to_page=100000,
             d = copy.deepcopy(doc)
             pn += from_page
             d["image"] = img
-            d["page_num_list"] = json.dumps([pn + 1])
-            d["top_list"] = json.dumps([0])
-            d["position_list"] = json.dumps([(pn + 1, 0, img.size[0], 0, img.size[1])])
+            d["page_num_int"] = [pn + 1]
+            d["top_int"] = [0]
+            d["position_int"] = [(pn + 1, 0, img.size[0], 0, img.size[1])]
             tokenize(d, txt, eng)
             res.append(d)
         return res
@@ -125,10 +124,9 @@ def chunk(filename, binary=None, from_page=0, to_page=100000,
             pn += from_page
             if img:
                 d["image"] = img
-            d["page_num_list"] = json.dumps([pn + 1])
-            d["top_list"] = json.dumps([0])
-            d["position_list"] = json.dumps([
-                (pn + 1, 0, img.size[0] if img else 0, 0, img.size[1] if img else 0)])
+            d["page_num_int"] = [pn + 1]
+            d["top_int"] = [0]
+            d["position_int"] = [(pn + 1, 0, img.size[0] if img else 0, 0, img.size[1] if img else 0)]
             tokenize(d, txt, eng)
             res.append(d)
         return res

rag/nlp/__init__.py

@@ -22,7 +22,6 @@ from rag.utils import num_tokens_from_string
 from . import rag_tokenizer
 import re
 import copy
-import json
 import roman_numbers as r
 from word2number import w2n
 from cn2an import cn2an
@@ -311,16 +310,16 @@ def tokenize_table(tbls, doc, eng, batch_size=10):
 def add_positions(d, poss):
     if not poss:
         return
-    page_num_list = []
-    position_list = []
-    top_list = []
+    page_num_int = []
+    position_int = []
+    top_int = []
     for pn, left, right, top, bottom in poss:
-        page_num_list.append(int(pn + 1))
-        top_list.append(int(top))
-        position_list.append((int(pn + 1), int(left), int(right), int(top), int(bottom)))
-    d["page_num_list"] = json.dumps(page_num_list)
-    d["position_list"] = json.dumps(position_list)
-    d["top_list"] = json.dumps(top_list)
+        page_num_int.append(int(pn + 1))
+        top_int.append(int(top))
+        position_int.append((int(pn + 1), int(left), int(right), int(top), int(bottom)))
+    d["page_num_int"] = page_num_int
+    d["position_int"] = position_int
+    d["top_int"] = top_int
 
 
 def remove_contents_table(sections, eng=False):

rag/nlp/search.py

@@ -16,7 +16,6 @@
 
 import logging
 import re
-import json
 from dataclasses import dataclass
 
 from rag.utils import rmSpace
@@ -74,7 +73,7 @@ class Dealer:
         offset, limit = pg * ps, (pg + 1) * ps
 
         src = req.get("fields", ["docnm_kwd", "content_ltks", "kb_id", "img_id", "title_tks", "important_kwd",
-                                 "doc_id", "position_list", "knowledge_graph_kwd", "question_kwd", "question_tks",
+                                 "doc_id", "page_num_int", "top_int", "create_timestamp_flt", "knowledge_graph_kwd", "question_kwd", "question_tks",
                                  "available_int", "content_with_weight", "pagerank_fea"])
         kwds = set([])
 
@@ -82,6 +81,8 @@ class Dealer:
         q_vec = []
         if not qst:
             if req.get("sort"):
+                orderBy.asc("page_num_int")
+                orderBy.asc("top_int")
                 orderBy.desc("create_timestamp_flt")
             res = self.dataStore.search(src, [], filters, [], orderBy, offset, limit, idx_names, kb_ids)
             total=self.dataStore.getTotal(res)
@@ -340,7 +341,7 @@ class Dealer:
             chunk = sres.field[id]
             dnm = chunk["docnm_kwd"]
             did = chunk["doc_id"]
-            position_list = chunk.get("position_list", "[]")
+            position_int = chunk.get("position_int", [])
             d = {
                 "chunk_id": id,
                 "content_ltks": chunk["content_ltks"],
@@ -354,7 +355,7 @@ class Dealer:
                 "vector_similarity": vsim[i],
                 "term_similarity": tsim[i],
                 "vector": chunk.get(vector_column, zero_vector),
-                "positions": json.loads(position_list)
+                "positions": position_int,
             }
             if highlight and sres.highlight:
                 if id in sres.highlight:

rag/svr/task_executor.py

@@ -211,9 +211,9 @@ def build_chunks(task, progress_callback):
         if not d.get("image"):
             _ = d.pop("image", None)
             d["img_id"] = ""
-            d["page_num_list"] = json.dumps([])
-            d["position_list"] = json.dumps([])
-            d["top_list"] = json.dumps([])
+            d["page_num_int"] = []
+            d["position_int"] = []
+            d["top_int"] = []
             docs.append(d)
             continue
 

rag/utils/es_conn.py

@@ -185,8 +185,14 @@ class ESConnection(DocStoreConnection):
             orders = list()
             for field, order in orderBy.fields:
                 order = "asc" if order == 0 else "desc"
-                orders.append({field: {"order": order, "unmapped_type": "float",
-                                       "mode": "avg", "numeric_type": "double"}})
+                if field in ["page_num_int", "top_int"]:
+                    order_info = {"order": order, "unmapped_type": "float",
+                                "mode": "avg", "numeric_type": "double"}
+                elif field.endswith("_int") or field.endswith("_flt"):
+                    order_info = {"order": order, "unmapped_type": "float"}
+                else:
+                    order_info = {"order": order, "unmapped_type": "text"}
+                orders.append({field: order_info})
             s = s.sort(*orders)
 
         if limit > 0:

rag/utils/infinity_conn.py

@@ -297,7 +297,7 @@ class InfinityConnection(DocStoreConnection):
                 df_list.append(kb_res)
         self.connPool.release_conn(inf_conn)
         res = concat_dataframes(df_list, selectFields)
-        logger.debug("INFINITY search tables: " + str(table_list))
+        logger.debug(f"INFINITY search tables: {str(table_list)}, result: {str(res)}")
         return res
 
     def get(
@@ -307,8 +307,10 @@ class InfinityConnection(DocStoreConnection):
         db_instance = inf_conn.get_database(self.dbName)
         df_list = list()
         assert isinstance(knowledgebaseIds, list)
+        table_list = list()
         for knowledgebaseId in knowledgebaseIds:
             table_name = f"{indexName}_{knowledgebaseId}"
+            table_list.append(table_name)
             table_instance = db_instance.get_table(table_name)
             kb_res = table_instance.output(["*"]).filter(f"id = '{chunkId}'").to_pl()
             if len(kb_res) != 0 and kb_res.shape[0] > 0:
@@ -316,6 +318,7 @@ class InfinityConnection(DocStoreConnection):
 
         self.connPool.release_conn(inf_conn)
         res = concat_dataframes(df_list, ["id"])
+        logger.debug(f"INFINITY get tables: {str(table_list)}, result: {str(res)}")
         res_fields = self.getFields(res, res.columns)
         return res_fields.get(chunkId, None)
 
@@ -349,15 +352,22 @@ class InfinityConnection(DocStoreConnection):
             for k, v in d.items():
                 if k.endswith("_kwd") and isinstance(v, list):
                     d[k] = " ".join(v)
-                if k == 'kb_id':
+                elif k == 'kb_id':
                     if isinstance(d[k], list):
                         d[k] = d[k][0] # since d[k] is a list, but we need a str
+                elif k == "position_int":
+                    assert isinstance(v, list)
+                    arr = [num for row in v for num in row]
+                    d[k] = "_".join(f"{num:08x}" for num in arr)
+                elif k in ["page_num_int", "top_int", "position_int"]:
+                    assert isinstance(v, list)
+                    d[k] = "_".join(f"{num:08x}" for num in v)
         ids = ["'{}'".format(d["id"]) for d in documents]
         str_ids = ", ".join(ids)
         str_filter = f"id IN ({str_ids})"
         table_instance.delete(str_filter)
         # for doc in documents:
-        #     logger.info(f"insert position_list: {doc['position_list']}")
+        #     logger.info(f"insert position_int: {doc['position_int']}")
         # logger.info(f"InfinityConnection.insert {json.dumps(documents)}")
         table_instance.insert(documents)
         self.connPool.release_conn(inf_conn)
@@ -367,8 +377,8 @@ class InfinityConnection(DocStoreConnection):
     def update(
             self, condition: dict, newValue: dict, indexName: str, knowledgebaseId: str
     ) -> bool:
-        # if 'position_list' in newValue:
-        #     logger.info(f"upsert position_list: {newValue['position_list']}")
+        # if 'position_int' in newValue:
+        #     logger.info(f"update position_int: {newValue['position_int']}")
         inf_conn = self.connPool.get_conn()
         db_instance = inf_conn.get_database(self.dbName)
         table_name = f"{indexName}_{knowledgebaseId}"
@@ -377,6 +387,16 @@ class InfinityConnection(DocStoreConnection):
         for k, v in newValue.items():
             if k.endswith("_kwd") and isinstance(v, list):
                 newValue[k] = " ".join(v)
+            elif k == 'kb_id':
+                if isinstance(newValue[k], list):
+                    newValue[k] = newValue[k][0] # since d[k] is a list, but we need a str
+            elif k == "position_int":
+                assert isinstance(v, list)
+                arr = [num for row in v for num in row]
+                newValue[k] = "_".join(f"{num:08x}" for num in arr)
+            elif k in ["page_num_int", "top_int"]:
+                assert isinstance(v, list)
+                newValue[k] = "_".join(f"{num:08x}" for num in v)
         table_instance.update(filter, newValue)
         self.connPool.release_conn(inf_conn)
         return True
@@ -423,9 +443,22 @@ class InfinityConnection(DocStoreConnection):
                 v = res[fieldnm][i]
                 if isinstance(v, Series):
                     v = list(v)
-                elif fieldnm == "important_kwd":
+                elif fieldnm.endswith("_kwd"):
                     assert isinstance(v, str)
                     v = v.split()
+                elif fieldnm == "position_int":
+                    assert isinstance(v, str)
+                    if v:
+                        arr = [int(hex_val, 16) for hex_val in v.split('_')]
+                        v = [arr[i:i + 4] for i in range(0, len(arr), 4)]
+                    else:
+                        v = []
+                elif fieldnm in ["page_num_int", "top_int"]:
+                    assert isinstance(v, str)
+                    if v:
+                        v = [int(hex_val, 16) for hex_val in v.split('_')]
+                    else:
+                        v = []
                 else:
                     if not isinstance(v, str):
                         v = str(v)