From dc312b0852b6d5cb30464a90b4e176528d34d124 Mon Sep 17 00:00:00 2001
From: HuangHai <10402852@qq.com>
Date: Fri, 27 Jun 2025 14:45:10 +0800
Subject: [PATCH] 'commit'

---
 dsRag/Util/EsSearchUtil.py                    |  22 ++++++++++++++----
 .../__pycache__/EsSearchUtil.cpython-310.pyc  | Bin 2627 -> 3156 bytes
 2 files changed, 17 insertions(+), 5 deletions(-)

diff --git a/dsRag/Util/EsSearchUtil.py b/dsRag/Util/EsSearchUtil.py
index d2aeece3..6ded7acc 100644
--- a/dsRag/Util/EsSearchUtil.py
+++ b/dsRag/Util/EsSearchUtil.py
@@ -1,6 +1,9 @@
-from elasticsearch import Elasticsearch
 import jieba
-import numpy as np
+from elasticsearch import Elasticsearch
+from gensim.models import KeyedVectors
+
+from Config.Config import MS_MODEL_PATH, MS_MODEL_LIMIT
+
 
 class EsSearchUtil:
     def __init__(self, es_config):
@@ -16,9 +19,18 @@ class EsSearchUtil:
         )
 
     def text_to_embedding(self, text):
-        # 当前实现为随机向量生成，后续可替换为实际模型
-        vector = np.random.rand(200).tolist()
-        return vector
+        # 加载预训练模型
+        model = KeyedVectors.load_word2vec_format(MS_MODEL_PATH, binary=False, limit=MS_MODEL_LIMIT)
+        # 对文本分词并计算平均向量
+        words = jieba.lcut(text)
+        vectors = [model[word] for word in words if word in model]
+        
+        if not vectors:
+            return [0.0] * model.vector_size
+            
+        # 计算平均向量
+        avg_vector = [sum(dim)/len(vectors) for dim in zip(*vectors)]
+        return avg_vector
 
     def vector_search(self, query, size=10):
         query_embedding = self.text_to_embedding(query)
diff --git a/dsRag/Util/__pycache__/EsSearchUtil.cpython-310.pyc b/dsRag/Util/__pycache__/EsSearchUtil.cpython-310.pyc
index b0a8cb27fee52a7b6c3b705436dac01b565a8512..f57645b76a53df43ffe44389e2a7afbeccdc4c35 100644
GIT binary patch
delta 1327
zcmZWp&u`R56rML@d!60&LbppOZA57}RHigQ_;FeyK}ZXNK%=D`R-)U@j<a!Oufy0$
zNwn!!3KHSaR+0aKR_dk4D%ak6?U6$@=U#g5xl-R7ha`Zdc{B6oy>Gtx-i-ffynL^g
zcwU*|_viJmmVc@2)+R<baVD9Bh_$Kb5yZUBfyN>dvDLQB*oy4fX*+SLT>|zwleTo8
zG3jjcwkx5<{L()Jd*27`%|+cb3wvqMRlQ|&RPP2GfqWG7vP7jm*WP@qGymY$+`Y~h
zHy+k+>)OHm-kte74|}`socis@2jG+tP`UZvaLpyozcS_Jll(f2<zMsW$VQjImOet0
zu@%U;giJ{Oy?DVg>nD2%Kf@M(&Xzz5X+3j?{1)4W7qA2AX=(Rihpjjb)?&{ABOHpz
zN;~iaqo^%06m+QjpCMmoF&uQ&hPI<H4znKMevFNbWgI&T+q>A(2{e5+WRJ^RFhu-y
zruYsU!YQ`PpYRs)#{gpxl}}yN);EcgsE{2eGKh-h8ph~yGYZqJm&B_x@8kSm+Gc`n
zPS2$+qw1Ggz91T-sb-Y)x=}iVgIDo=3^m_A(PBaXq}W?LGvv1o8M57m503IgiXvT$
zwTOa&&s80hwUA+qbW7_+Ek}op)d^fbgH4U2@m)ow(TynS$_^cJ8dEw;iHf_KBE1xm
z(e_FhEOy&g)LYARr9kbZ;bx%4X1MBGs!CB-7Cg<gO)zP@++FW?ibz!_J&{|307)lR
z%g>1M6T~GY9gp)euY<*_FfR6H#cjCsF?zK{TOj&Jlu{v&Q(RU>_sqA=wu(a4`TTe5
z?CFELDN`Nq6;o4-v-yA4ci`vW+9zkvQNA#k>Ql-$fZ`}1dEdaR&;dIIg{ymi*^goN
z0)<Tz+G}jcdzs7?GH<|m%NvU-l>0!<y{DxCENEw6tPcrHr~_bil^lh{=B&C>Fz93(
ztARTI_6Fueck?S_pLmm$%hfdsn#`lIbE9K4G=jdqLm|T<Q!^qJm4n8`W-8?}7u;&O
zR<4&zwPR+@A5-+8v=<!xpD4P?+Uo~{G>oUryGik~Es%3I87zhUsiHf)W(MtuSO*le
X+eCo1D43x)ol>n_R)s1Y=XLQv%6b|k

delta 845
zcmZuv&ubGw6yBNL*=#o1*oI;wXqA+L+e)YxFIGJi(SyW;w}^&hXKA+VkMMS4jkHuy
z@LcvENDm?&C4WLMg4dqrFR=fG^JYcc>cYHv^UXKk_ulNBwoaBBndiBN`aa!%J^E2S
zYFvNTQ7dk65*rEW5rR35X^-mJ>{(9?ZgT6R!L1Lp$2fgq`0NZNd?LNa@gNsbkcR^Z
zM&00O@!Gf3=ft+uHn75{_neAKzZ&2cJM;mui_diH&{WX28u^L}!kCeLa!d}Wpu6VG
z;wqo9W8=WwH<cCbo_$v=-<E7Q3Woz}#lb{KnvSIngOq29v~ZTT$l@p$z6DjqmzD_M
z3d!~|;JLKlgn`Hauyb&4W2_-*5lMZK^}}Qsavr5SO$Fz+3NBe7^A3AT?%qJAl}GPF
zrM;{c;2>g`Jp=aVGi?e;g?919UR^1zaIKxZvVwkjXu0@lzkhiXWmqy0!AQE3JcNFf
zjwb?cDcm=;3#_7owlL3IAbQE6rh6&%{rO4d71>@x8g_fJK16*C>liM{c%$hMBL2J3
z&3RBwNf)jnb4dkN*B@2bK%l&D-MXxm_I)uOhtU34UAOb7XxBEp7M`VW2T84>S6e@<
zW2*1>qC<HHf6LSvUMk=_Vrt|K6&%AMPTgg0)2%eF=soplyzu-31&YUzEm=BA##2~8
c0Ck}A9|!4dB^jT@;X|S*IsW%L+K?vw1wxy#od5s;