AI/ragflow
1
0
Fork 0
mirror of https://git.mirrors.martin98.com/https://github.com/infiniflow/ragflow.git synced 2025-08-11 13:59:03 +08:00
Code Issues Packages Projects Releases Wiki Activity

Update progress info and start welcome info (#3768)

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

_Briefly describe what this PR aims to solve. Include background context
that will help reviewers understand the purpose of the PR._

### Type of change

- [x] Refactoring

---------

Signed-off-by: jinhai <haijin.chn@gmail.com>
This commit is contained in:
Jin Hai 2024-11-30 18:48:06 +08:00 committed by GitHub
parent d00297a763
commit e079656473
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
11 changed files with 107 additions and 59 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
api/ragflow_server.py
View File

@ -47,6 +47,7 @@ from api.db.db_models import init_database_tables as init_web_db
from api.db.init_data import init_web_data
from api.versions import get_ragflow_version
from api.utils import show_configs
from rag.settings import print_rag_settings
def update_progress():
@ -75,6 +76,7 @@ if __name__ == '__main__':
)
show_configs()
settings.init_settings()
print_rag_settings()
# init db
init_web_db()

19
rag/app/book.py
View File

@ -26,30 +26,33 @@ from deepdoc.parser import PdfParser, DocxParser, PlainParser, HtmlParser
class Pdf(PdfParser):
def __call__(self, filename, binary=None, from_page=0,
to_page=100000, zoomin=3, callback=None):
callback(msg="OCR is running...")
from timeit import default_timer as timer
start = timer()
callback(msg="OCR started")
self.__images__(
filename if not binary else binary,
zoomin,
from_page,
to_page,
callback)
callback(msg="OCR finished")
callback(msg="OCR finished ({:.2f}s)".format(timer() - start))
from timeit import default_timer as timer
start = timer()
self._layouts_rec(zoomin)
callback(0.67, "Layout analysis finished")
callback(0.67, "Layout analysis ({:.2f}s)".format(timer() - start))
logging.debug("layouts: {}".format(timer() - start))
start = timer()
self._table_transformer_job(zoomin)
callback(0.68, "Table analysis finished")
callback(0.68, "Table analysis ({:.2f}s)".format(timer() - start))
start = timer()
self._text_merge()
tbls = self._extract_table_figure(True, zoomin, True, True)
self._naive_vertical_merge()
self._filter_forpages()
self._merge_with_same_bullet()
callback(0.75, "Text merging finished.")
callback(0.8, "Text extraction finished")
callback(0.8, "Text extraction ({:.2f}s)".format(timer() - start))
return [(b["text"] + self._line_tag(b, zoomin), b.get("layoutno", ""))
for b in self.boxes], tbls

11
rag/app/laws.py
View File

@ -108,7 +108,9 @@ class Pdf(PdfParser):
def __call__(self, filename, binary=None, from_page=0,
to_page=100000, zoomin=3, callback=None):
callback(msg="OCR is running...")
from timeit import default_timer as timer
start = timer()
callback(msg="OCR started")
self.__images__(
filename if not binary else binary,
zoomin,
@ -116,17 +118,16 @@ class Pdf(PdfParser):
to_page,
callback
)
callback(msg="OCR finished")
callback(msg="OCR finished ({:.2f}s)".format(timer() - start))
from timeit import default_timer as timer
start = timer()
self._layouts_rec(zoomin)
callback(0.67, "Layout analysis finished")
callback(0.67, "Layout analysis ({:.2f}s)".format(timer() - start))
logging.debug("layouts:".format(
))
self._naive_vertical_merge()
callback(0.8, "Text extraction finished")
callback(0.8, "Text extraction ({:.2f}s)".format(timer() - start))
return [(b["text"], self._line_tag(b, zoomin))
for b in self.boxes], None

15
rag/app/manual.py
View File

@ -36,7 +36,7 @@ class Pdf(PdfParser):
to_page=100000, zoomin=3, callback=None):
from timeit import default_timer as timer
start = timer()
callback(msg="OCR is running...")
callback(msg="OCR started")
self.__images__(
filename if not binary else binary,
zoomin,
@ -44,22 +44,27 @@ class Pdf(PdfParser):
to_page,
callback
)
callback(msg="OCR finished.")
callback(msg="OCR finished ({:.2f}s)".format(timer() - start))
# for bb in self.boxes:
# for b in bb:
# print(b)
logging.debug("OCR: {}".format(timer() - start))
start = timer()
self._layouts_rec(zoomin)
callback(0.65, "Layout analysis finished.")
callback(0.65, "Layout analysis ({:.2f}s)".format(timer() - start))
logging.debug("layouts: {}".format(timer() - start))
start = timer()
self._table_transformer_job(zoomin)
callback(0.67, "Table analysis finished.")
callback(0.67, "Table analysis ({:.2f}s)".format(timer() - start))
start = timer()
self._text_merge()
tbls = self._extract_table_figure(True, zoomin, True, True)
self._concat_downward()
self._filter_forpages()
callback(0.68, "Text merging finished")
callback(0.68, "Text merged ({:.2f}s)".format(timer() - start))
# clean mess
for b in self.boxes:

21
rag/app/naive.py
View File

@ -124,7 +124,8 @@ class Pdf(PdfParser):
def __call__(self, filename, binary=None, from_page=0,
to_page=100000, zoomin=3, callback=None):
start = timer()
callback(msg="OCR is running...")
first_start = start
callback(msg="OCR started")
self.__images__(
filename if not binary else binary,
zoomin,
@ -132,22 +133,26 @@ class Pdf(PdfParser):
to_page,
callback
)
callback(msg="OCR finished")
logging.info("OCR({}~{}): {}".format(from_page, to_page, timer() - start))
callback(msg="OCR finished ({:.2f}s)".format(timer() - start))
logging.info("OCR({}~{}): {:.2f}s".format(from_page, to_page, timer() - start))
start = timer()
self._layouts_rec(zoomin)
callback(0.63, "Layout analysis finished.")
callback(0.63, "Layout analysis ({:.2f}s)".format(timer() - start))
start = timer()
self._table_transformer_job(zoomin)
callback(0.65, "Table analysis finished.")
callback(0.65, "Table analysis ({:.2f}s)".format(timer() - start))
start = timer()
self._text_merge()
callback(0.67, "Text merging finished")
callback(0.67, "Text merged ({:.2f}s)".format(timer() - start))
tbls = self._extract_table_figure(True, zoomin, True, True)
# self._naive_vertical_merge()
self._concat_downward()
# self._filter_forpages()
logging.info("layouts cost: {}s".format(timer() - start))
logging.info("layouts cost: {}s".format(timer() - first_start))
return [(b["text"], self._line_tag(b, zoomin))
for b in self.boxes], tbls
@ -170,7 +175,7 @@ class Markdown(MarkdownParser):
else:
if sections and sections[-1][0].strip().find("#") == 0:
sec_, _ = sections.pop(-1)
sections.append((sec_+"\n"+sec, ""))
sections.append((sec_ + "\n" + sec, ""))
else:
sections.append((sec, ""))

17
rag/app/one.py
View File

@ -24,7 +24,9 @@ from deepdoc.parser import PdfParser, ExcelParser, PlainParser, HtmlParser
class Pdf(PdfParser):
def __call__(self, filename, binary=None, from_page=0,
to_page=100000, zoomin=3, callback=None):
callback(msg="OCR is running...")
from timeit import default_timer as timer
start = timer()
callback(msg="OCR started")
self.__images__(
filename if not binary else binary,
zoomin,
@ -32,17 +34,20 @@ class Pdf(PdfParser):
to_page,
callback
)
callback(msg="OCR finished")
callback(msg="OCR finished ({:.2f}s)".format(timer() - start))
from timeit import default_timer as timer
start = timer()
self._layouts_rec(zoomin, drop=False)
callback(0.63, "Layout analysis finished.")
callback(0.63, "Layout analysis ({:.2f}s)".format(timer() - start))
logging.debug("layouts cost: {}s".format(timer() - start))
start = timer()
self._table_transformer_job(zoomin)
callback(0.65, "Table analysis finished.")
callback(0.65, "Table analysis ({:.2f}s)".format(timer() - start))
start = timer()
self._text_merge()
callback(0.67, "Text merging finished")
callback(0.67, "Text merged ({:.2f}s)".format(timer() - start))
tbls = self._extract_table_figure(True, zoomin, True, True)
self._concat_downward()

17
rag/app/paper.py
View File

@ -27,7 +27,9 @@ class Pdf(PdfParser):
def __call__(self, filename, binary=None, from_page=0,
to_page=100000, zoomin=3, callback=None):
callback(msg="OCR is running...")
from timeit import default_timer as timer
start = timer()
callback(msg="OCR started")
self.__images__(
filename if not binary else binary,
zoomin,
@ -35,21 +37,24 @@ class Pdf(PdfParser):
to_page,
callback
)
callback(msg="OCR finished.")
callback(msg="OCR finished ({:.2f}s)".format(timer() - start))
from timeit import default_timer as timer
start = timer()
self._layouts_rec(zoomin)
callback(0.63, "Layout analysis finished")
callback(0.63, "Layout analysis ({:.2f}s)".format(timer() - start))
logging.debug(f"layouts cost: {timer() - start}s")
start = timer()
self._table_transformer_job(zoomin)
callback(0.68, "Table analysis finished")
callback(0.68, "Table analysis ({:.2f}s)".format(timer() - start))
start = timer()
self._text_merge()
tbls = self._extract_table_figure(True, zoomin, True, True)
column_width = np.median([b["x1"] - b["x0"] for b in self.boxes])
self._concat_downward()
self._filter_forpages()
callback(0.75, "Text merging finished.")
callback(0.75, "Text merged ({:.2f}s)".format(timer() - start))
# clean mess
if column_width < self.page_images[0].size[0] / zoomin / 2:

7
rag/app/presentation.py
View File

@ -59,11 +59,12 @@ class Pdf(PdfParser):
def __call__(self, filename, binary=None, from_page=0,
to_page=100000, zoomin=3, callback=None):
callback(msg="OCR is running...")
from timeit import default_timer as timer
start = timer()
callback(msg="OCR started")
self.__images__(filename if not binary else binary,
zoomin, from_page, to_page, callback)
callback(0.8, "Page {}~{}: OCR finished".format(
from_page, min(to_page, self.total_page)))
callback(msg="Page {}~{}: OCR finished ({:.2f}s)".format(from_page, min(to_page, self.total_page), timer() - start))
assert len(self.boxes) == len(self.page_images), "{} vs. {}".format(
len(self.boxes), len(self.page_images))
res = []

18
rag/app/qa.py
View File

@ -73,7 +73,7 @@ class Pdf(PdfParser):
def __call__(self, filename, binary=None, from_page=0,
to_page=100000, zoomin=3, callback=None):
start = timer()
callback(msg="OCR is running...")
callback(msg="OCR started")
self.__images__(
filename if not binary else binary,
zoomin,
@ -81,15 +81,19 @@ class Pdf(PdfParser):
to_page,
callback
)
callback(msg="OCR finished")
logging.debug("OCR({}~{}): {}".format(from_page, to_page, timer() - start))
callback(msg="OCR finished ({:.2f}s)".format(timer() - start))
logging.debug("OCR({}~{}): {:.2f}s".format(from_page, to_page, timer() - start))
start = timer()
self._layouts_rec(zoomin, drop=False)
callback(0.63, "Layout analysis finished.")
callback(0.63, "Layout analysis ({:.2f}s)".format(timer() - start))
start = timer()
self._table_transformer_job(zoomin)
callback(0.65, "Table analysis finished.")
callback(0.65, "Table analysis ({:.2f}s)".format(timer() - start))
start = timer()
self._text_merge()
callback(0.67, "Text merging finished")
callback(0.67, "Text merged ({:.2f}s)".format(timer() - start))
tbls = self._extract_table_figure(True, zoomin, True, True)
#self._naive_vertical_merge()
# self._concat_downward()
@ -226,7 +230,7 @@ class Docx(DocxParser):
sum_question = '\n'.join(question_stack)
if sum_question:
qai_list.append((sum_question, last_answer, last_image))
tbls = []
for tb in self.doc.tables:
html= "<table>"

7
rag/settings.py
View File

@ -14,6 +14,7 @@
# limitations under the License.
#
import os
import logging
from api.utils import get_base_config, decrypt_database_config
from api.utils.file_utils import get_project_base_directory
@ -37,3 +38,9 @@ SVR_QUEUE_RETENTION = 60*60
SVR_QUEUE_MAX_LEN = 1024
SVR_CONSUMER_NAME = "rag_flow_svr_consumer"
SVR_CONSUMER_GROUP_NAME = "rag_flow_svr_consumer_group"
def print_rag_settings():
logging.info(f"MAX_CONTENT_LENGTH: {DOC_MAXIMUM_SIZE}")
logging.info(f"SERVER_QUEUE_MAX_LEN: {SVR_QUEUE_MAX_LEN}")
logging.info(f"SERVER_QUEUE_RETENTION: {SVR_QUEUE_RETENTION}")
logging.info(f"MAX_FILE_COUNT_PER_USER: {int(os.environ.get('MAX_FILE_NUM_PER_USER', 0))}")

32
rag/svr/task_executor.py
View File

@ -56,12 +56,13 @@ from api.db.services.llm_service import LLMBundle
from api.db.services.task_service import TaskService
from api.db.services.file2document_service import File2DocumentService
from api import settings
from api.versions import get_ragflow_version
from api.db.db_models import close_connection
from rag.app import laws, paper, presentation, manual, qa, table, book, resume, picture, naive, one, audio, \
knowledge_graph, email
from rag.nlp import search, rag_tokenizer
from rag.raptor import RecursiveAbstractiveProcessing4TreeOrganizedRetrieval as Raptor
from rag.settings import DOC_MAXIMUM_SIZE, SVR_QUEUE_NAME
from rag.settings import DOC_MAXIMUM_SIZE, SVR_QUEUE_NAME, print_rag_settings
from rag.utils import rmSpace, num_tokens_from_string
from rag.utils.redis_conn import REDIS_CONN, Payload
from rag.utils.storage_factory import STORAGE_IMPL
@ -395,8 +396,7 @@ def do_handle_task(r):
# TODO: exception handler
## set_progress(r["did"], -1, "ERROR: ")
callback(
msg="Finished slicing files ({} chunks in {:.2f}s). Start to embedding the content.".format(len(cks),
timer() - st)
msg="Generate {} chunks ({:.2f}s). Embedding chunks.".format(len(cks), timer() - st)
)
st = timer()
try:
@ -407,7 +407,7 @@ def do_handle_task(r):
tk_count = 0
raise
logging.info("Embedding elapsed({}): {:.2f}".format(r["name"], timer() - st))
callback(msg="Finished embedding (in {:.2f}s)! Start to build index!".format(timer() - st))
callback(msg="Finished embedding ({:.2f}s)!".format(timer() - st))
# logging.info(f"task_executor init_kb index {search.index_name(r["tenant_id"])} embd_mdl {embd_mdl.llm_name} vector length {vector_size}")
init_kb(r, vector_size)
chunk_count = len(set([c["id"] for c in cks]))
@ -420,7 +420,8 @@ def do_handle_task(r):
callback(prog=0.8 + 0.1 * (b + 1) / len(cks), msg="")
logging.info("Indexing elapsed({}): {:.2f}".format(r["name"], timer() - st))
if es_r:
callback(-1, "Insert chunk error, detail info please check log file. Please also check Elasticsearch/Infinity status!")
callback(-1,
"Insert chunk error, detail info please check log file. Please also check Elasticsearch/Infinity status!")
settings.docStoreConn.delete({"doc_id": r["doc_id"]}, search.index_name(r["tenant_id"]), r["kb_id"])
logging.error('Insert chunk error: ' + str(es_r))
raise Exception('Insert chunk error: ' + str(es_r))
@ -429,13 +430,12 @@ def do_handle_task(r):
settings.docStoreConn.delete({"doc_id": r["doc_id"]}, search.index_name(r["tenant_id"]), r["kb_id"])
return
callback(msg="Indexing elapsed in {:.2f}s.".format(timer() - st))
callback(1., "Done!")
callback(1., msg="Index cost {:.2f}s.".format(timer() - st))
DocumentService.increment_chunk_num(
r["doc_id"], r["kb_id"], tk_count, chunk_count, 0)
logging.info(
"Chunk doc({}), token({}), chunks({}), elapsed:{:.2f}".format(
r["id"], tk_count, len(cks), timer() - st))
r["id"], tk_count, len(cks), timer() - st))
def handle_task():
@ -502,7 +502,7 @@ def analyze_heap(snapshot1: tracemalloc.Snapshot, snapshot2: tracemalloc.Snapsho
for stat in stats2[:10]:
msg += f"{stat}\n"
stats1_vs_2 = snapshot2.compare_to(snapshot1, 'lineno')
msg += f"{CONSUMER_NAME} memory usage increase from snapshot {snapshot_id-1} to snapshot {snapshot_id}:\n"
msg += f"{CONSUMER_NAME} memory usage increase from snapshot {snapshot_id - 1} to snapshot {snapshot_id}:\n"
for stat in stats1_vs_2[:10]:
msg += f"{stat}\n"
msg += f"{CONSUMER_NAME} detailed traceback for the top memory consumers:\n"
@ -512,7 +512,16 @@ def analyze_heap(snapshot1: tracemalloc.Snapshot, snapshot2: tracemalloc.Snapsho
def main():
logging.info(r"""
______ __ ______ __
/_ __/___ ______/ /__ / ____/ _____ _______ __/ /_____ _____
/ / / __ `/ ___/ //_/ / __/ | |/_/ _ \/ ___/ / / / __/ __ \/ ___/
/ / / /_/ (__ ) ,< / /____> </ __/ /__/ /_/ / /_/ /_/ / /
/_/ \__,_/____/_/|_| /_____/_/|_|\___/\___/\__,_/\__/\____/_/
""")
logging.info(f'TaskExecutor: RAGFlow version: {get_ragflow_version()}')
settings.init_settings()
print_rag_settings()
background_thread = threading.Thread(target=report_status)
background_thread.daemon = True
background_thread.start()
@ -527,11 +536,12 @@ def main():
while True:
handle_task()
num_tasks = DONE_TASKS + FAILED_TASKS
if TRACE_MALLOC_DELTA> 0 and num_tasks > 0 and num_tasks % TRACE_MALLOC_DELTA == 0:
if TRACE_MALLOC_DELTA > 0 and num_tasks > 0 and num_tasks % TRACE_MALLOC_DELTA == 0:
snapshot2 = tracemalloc.take_snapshot()
analyze_heap(snapshot1, snapshot2, int(num_tasks/TRACE_MALLOC_DELTA), num_tasks % TRACE_MALLOC_FULL == 0)
analyze_heap(snapshot1, snapshot2, int(num_tasks / TRACE_MALLOC_DELTA), num_tasks % TRACE_MALLOC_FULL == 0)
snapshot1 = snapshot2
snapshot2 = None
if __name__ == "__main__":
main()