From 9d7109b60f4f3834162b1554ccff5ea58574291e Mon Sep 17 00:00:00 2001
From: Narayanan Madaswamy <maddy23285@gmail.com>
Date: Thu, 22 Jan 2026 21:08:50 +0530
Subject: [PATCH] structured data extraction from pdf

---
 backend/__pycache__/main.cpython-313.pyc | Bin 12009 -> 12724 bytes
 backend/main.py                          |  45 ++++++++++++++++-------
 backend/requirements.txt                 |   1 +
 frontend/src/app/ocr/ocr.component.ts    |   2 +-
 4 files changed, 34 insertions(+), 14 deletions(-)

diff --git a/backend/__pycache__/main.cpython-313.pyc b/backend/__pycache__/main.cpython-313.pyc
index 200e12e9f5aa7c61b6f0bee73d7b7754a66dd922..c17dccd75df8480809ec6d14536f528edf974d20 100644
GIT binary patch
delta 3887
zcma)8du&tJ89&$eUdNBv*h!3?#39Z*&Vvx*zygG4pwuCZ=7I*rk!9j*YVaeRdoz{@
z%^htssJ0S%(5h}l&8A9Y+E&oAO01P8+OqvwC4-IVj%xj*(l)78Q-Oq3wsqfkZXhAb
zq}jT^^WE=pzQ_5_IrpDKza4hpbGaM@e!tjB)92P)bO(gH7n?2>g#G{xC`57emq#mT
zMTjWQSfx@N3q~ZxRY6Wjv<jr|{x#8RS{)715NHr(Rew#imexk=XdQp|^w&olXhVp^
ztco{gi&&HrwCzTR@S}|}TP=w-Dy2OFZR#LO*)sxdjx`M`EF}2S&5BQ1-68PD-y?@e
zfYcHtP(hUPn5SD@n4^VrD}Y<MgxkuwLEu&`;kGTwVhzut9lQ;(Yl2T`V}BG{3muS!
z(7%qZ<6|f7gt5MVeYA^qE%<MMIUASE8R4FqfZM!;+s(OQ;I=H`Zs6Qj;I_pgk2BZ^
zF`P29DVT$aN_%XB66&!)2y0`-P^ju|!IVTqr6bm(tb?aBRuQo(>oJ)V!sEhjTAcEU
z8M|T|amHn{2})#PCcmU~FPmyJPQ|QZP3xljMFUCMxJ-Ww*B=z?Iu`XU$|i8IIo7)n
z?v`aXTft_Cy(Na_ZQS}#Vp!PD{w4mNeQK+5S|J5o?R(cO?6UK4vxVdYMaWqW34^w!
z*_m`H*;`9UPCO*C8}_n@Vo}5%AtxP@niq7F7D5Ip2{~M^NbSIDcY(TPpq>yX36VpV
zK^ME`uC*n~$)lHwh=z**|8B6wUx?I`HJr1Qw}u=oHRXgHm6%W)WLxb%yOjrP9kj8S
zy^6hWuVk0(e*aS64Nv&4U?102HEF$KHAx9I<f!}6oJAxNfLP_D-bWloX8*AV*6(Ku
zuy@EQAw~W#Gb@<2HWuwS$ic8+STl*S{HSgSj`5O_WNIWcnmwk{kV<KuZVIt=j@6|O
z!?7qar2L4QWBVMNw^Aq`2^Ud}$PhurkaYDsx*?4uj;n?pG0qu+YFMe7NUBsT0^Jb*
zu`#)w-EjB|O~As^Y0c2#&y#t{AvfQ3`7gv@j=wSZ=CeP3_KM?HRo87-<hnRl>O0pr
z7pQ#G{)T;0nhS>R1lwnW?bi;@1lwnVn<njlD-BNe%#{VMNVCDV*|N4*`@Sq9<+bxR
z;wif@et!JIne%7n{FSebUmRy=9;_nn?pe1hGG9zQ!3QpQn-8#mIKEYQ&;<0KU6F_P
zJ*3!oj|dQv%{Akxo$8#bn{hYZu{F)wnjX%JAb5B`z;!Hg3D>Ec+A-BN<8HcbYreb6
z`*2<c!B<)w5<J_|=O({u+|?_-zq$kP1KAFY4?H%+@*Q$t7x}QZ0&v5Jy^cO-(aj<e
zBsZNL-HyI>;?01&uT{L+DkI(H*zFT<_PTd_#6NgspxL|9)y5WZ6c%U)78Fm65N!lB
zY?*XQ)kYFIx*lF5>>as9SU>%V>=uLr><eol(hKZ+@W8>0(DMim0HGBFe4os@y)O&^
zObpOHT+F^{D`!`&o$RxsRR<3O%aHZ_Ni}B_WmQds<)=97e!3O4ku>eNgT45QfOd}@
zINdKq=UkqNeWv1Ga4_)HVbrWa_}{Y)pcny_d6Pl74%fuf=0qr&e#u!U_{)F_+bM1<
z3j2vr*y)5;up7m-N)?i85UK&flG)gDI*Xa`0N|V;ngoCyY!=oJ9ZwU(I`8y&isrL%
z(?9{Wq3Led7Z#z6HF!Q3YFMea$%@@htJrpLSp-W<pF!YGcOivV)Q*5LL!^gbFp7rr
zIbF@^bcmhvR<{kJybgf}gz6X&|C!4dL?J?I`j*#Y5hmEDrMrfC{ldj`8%nn$>_FHF
zV2Ej$z7JpVxTP43VM``-RZnMCLpqkvXJ`&3yAhTx=5gd9Ot6>B`mDS=j`QyLyeyh}
z1*K_(ApoOzA(^;7HnPwmVcSYE(LCDln0Reg@~oi>!Yk~uuhR6aanE;r(E>w%a#~lj
z8n&jM$e29@bzPwb6jzp~VLIjRS8{q9^$>7=4Ua$QHs`8I+xe}xr1CkZdm_sF&O%S~
zP~KZTP<Rcru#Cwu!<o?a#PD-jSWs;t9><FDJYESi9v&V~c%`7ufCGwvo-{uIZh?}?
zuJi4;B!9`A*EgA+_16DIsyFw<nd#U4do2aNE+^3TBtixui-3zAw$L%8$C36L@~Kgk
zj?pl@8S-#`G^ZQ3JcVODer$|e8d6f#h7D^zmC{t5o<&75k;#}vUFM2=IdTf;FCbt+
z;Wqgm(r8IvKtR`qRm~>S8I89)k7&s~vak;w$5lPPp!Nf>`N{O}DtaZ)5g-p5MWeZs
zx%}xIEo0W|E;DEBMLG%q&9^MFCyPM8i<%f>*hgPNnis&DN@sMqCehZAQ)#N{=ElXm
z4C`<{n@#IxX1wj~S=dO}@*2BdUDAi14QpZqF4H7$V7_*A*-G&0pnoXL4`JXnF=U=6
zEiiYeT4-czLw@-(2;ZKLh9ZKnXSz@`Vewo*^+^P*0X@uo^^@Vp;Ztl~Gw(}C*%A0M
zEF6d*)vpzX*uU!oB`S!*)<yU9b=K6-(t@YB5lBK+Gx<a^4yznr+Sk7Z@ojjtzY}&K
zv~Bv`h9j0B#x!2G<8k|ka(O+Jf<{WxBdnu2P?$mm=jIU9R?Q`{YA^jMaNx+oz>VU*
zor|_BW`MCS9O+11LBLA$lP4_Gw~@Yzz?<tgKpD<tS{uo0X+529&1o{Q77tRRbTPrl
zQPX!o=QH+Y^Q*!h_F_2N*@+Xj1K{@*RkhKKzVz;OYGb+KN4<F+)V^Y0hXY_6Xjv=t
zv!_}@dwX#P4*cxowYB3XQ=I1WI(-ksU&Euh0iGwH6#HiUT{F(ej1=LUIlRM6dW2nV
zIUB-z#SjlXy^mjW@FA8=tvBzaRUn~$wy(8Zk&lh(s>aWGigyw}>=VZlno99w+3az<
zHvF8L9L+$Rw<5C*6^n5%#zF7f>0y<wMLvF0&<^(V)~Z4elKgTXKuQ6C55hcb4t^)S
zh0Grz)0;|YdSWEK778|!&uI`0Up;=8@KX*C1imW_5jMAZ8{pQa*z6P=jPm_~%R#ZO
zP$#I|@;chKWb>HAR{9a>2=Hhp*mzr8<zI>SV^TZsC>Khm)OiBKHSs<U55B>E+4iE~
NnBLX?GeN=?`7iKTWdZ;I

delta 3496
zcmaJ@Yit}>6`tAm+q-_Qy-vJYJGR&FXB<0m(ztGD*0Bj00wP|tOm=6TO?GECXT~i~
zqOM6Q7bB?RUg!@{s}u=U0;E=Kk@y9Ke+a1pr9x(ae@IY7g{vTe;1B2ASv$s99cjLM
z=iGDddE9f(o%gaIelheU81xDFl-|Boc%ywQ6cs<%>iK5Hw-8OWLTg-*1j)b9mWn|v
z)ei0aJFsvhbrg=KI-rAp2N&Y0V{j}kq+L=d?U}Gib#1~8$0Tiw4<yo_gpf{1;b{>%
zhXpC}84<eDotGD^_5GL*rMskhsi86~N)7lnPCMg5R7eO?Q=1?)r^AzWiygW-wFRls
zebgRKZAEI^K5DNNODCoFbl-$SIzpyF|31o5P8mRhmiJ6yGZCM?Bi+m17yBzi7z1%(
z7>4;c0wXwfEsUndV60}`aa7p7PvJN>qX((I`=}F~nnY^fKI$Z=_9Jy5J-+AO2{cEu
z%wbcJ;Mu|Sq!gdFq2{4<0HYb36s^?>yEL3WDUIMenr@wNNn_;B4RJ5})3g&CjWkS+
zr%%ux2joR*q9*H*CJ)HY?8;7`*wyz5(o+YdPwh&xl07{9)MA@7gBnkzPuDCyeL!v&
z<(A1}_A`64^UU5ZAK9D5v&`Z6DLe0bC)|MkCq;<jpi?YZ!82IrDHigIUY1p8M^Y1u
z`TNB7>Un=i6i3;WKxLu_$xrYVAV$!+3!^yE0|>7O+o8z)MTCt-7~#w8E74Zg6d270
zFKR0V^)f2ZR9*RyLTow>?WUrd?vkSG@`?haU<rCjkfzoaT-O+V5|G@kE`R8LYC9O-
zxM(qsab52RCo0p(WI7Dx%f=^cIYnC%5V!6|K=XaUjSJRpv|{@5g`$$Vwq_{0=>o-A
z1vOjC9b6U!C%#lvPnwbXE#auJOsf%m1!3972ZYNG_HjoOTXfaCMM+d`S46fIYGk)v
zerD@vXK%Thn{0!~Hy9KI)wXzC5d1=1xME|sBJGJT;h3;Y!R3!zknzw|=W^XHMY6L2
zca**AZs>F@nSv?$ZZ|v!pg|lIIag3u;+N*n&%~1+z}*G5<31%PT`+@l@<`4p09Lrl
zcmk%EFhh|uMMZZZSDZim#LXs(sGK!8LnaSeDO1knA%$$S`06hqtP49fHsk#~n+b<b
zc0N4!aQxSyo)0{|f2fQ6XUA!C*8f{Sifs4JInQ+qkNj<f-M%@G{n1Eh&TjvO-HG$+
zg;1v$r9G2g2;pElw6dbYS)9T%?0#KBdX}d11kWKP9hOVZLWvTIJc0`dx&whcupDUl
ze+MjqOURb0{;TfqHgT5yrT!yvn*Fw+XNf!mu8HDpdKGoGq=0NFut+OQ1l$HXKA=s&
zOF+qkmS4ufbZ0fyP*el1vW3Quflt%=^90;JVki6fW7X9hEfPr8*Biq&QD<*AKbz&X
zkqkhRR{IG02?h{MdjYo}qEAXjAY)8V4$U+QCB<}H)3hSoq?KWU1ErNC8iCFhTjpK7
zY;(Mn-fc<c@6c+V;3|R{s5z4{*2=Yd#mqXyCMc2&w~5E*FwJFRA-Kc76pdM$%cwb9
zSx2JjEEq~jrzT;@MXQi8ribaGL*_163c2dT6dJ@wu#T@zWr`*5uY2IxJoDJmy6q2b
zq<Hb$z~BLRsdcgPGICJ}b8DtwHVip?y@ZFU*X;2fYR}#&SoU!9xU+{DwT23SfHdhX
z2-Jt_AB=AfK6bRQR9nD`%vv?w_Pnja&tipSt`igpZV=D`CvETo&0nN>#B|<TRbUP3
z@ym2(wN=$HJsM!iV@u<brX#25S<|KE^SWX{g&6H}v1kSJfGV!#3o7k@mw*D9bORMV
zmxOhik!sVWl;lED=PAxD;{A1hmYcp6#mLmyHc{rQ)w@S#9pTU6^s}aWRlTWdFRC!h
zetB%%@|p4v009&3fMx$11mHPKCMKKU9!c>4xUiRF;c{8inJ<8DSj9?ynXar>DisXN
zGoI?+5*81oxz6q+>gP$b>5|LXhH^Y<`I*6ili<-I{TP_9k?M64b8<i7F8fWQLyWPH
z6D`hfBmFzoVCRG=j#THmHf-Uq6Z<y^C<btaZFfIN?wJn{h{*lkqM4U~A49DtZuJa_
zSJ}^dqV?Bs30wT0dU&1vwWqHuhGa9E!>ATDIhVmx&g{?U@8j~L>dD?WZ56WQcH_Cd
zWRI(w5zk}#<hW_3tDqJOs$%+{%EKEZ<3AOr3saR#$}GHz6zo^zff<;ed%E4Bu8`5S
z_Kx?T0N*2^nDf4qbixm4{zHOwf_K>M{#Xb1so|+ldnUt=km+~qz5dt45%#HpRH}`H
znh}yB-Fy0K(cphRwTr>8uc_JHbo?>0{TW}q4q;vRQ?T=K@c8=qZGU7Pe{0z92ZqE+
z<{ON6@lMKN1UaoM#!HuK1~koppCIvXY-zA1NXyA*EdNRj2JclCkz(2}UASmnZ(h7d
z$E>S}l$s&^R#7fyuPddsAq`3pCL*<T`I@XNK&{KlYOjt=xz!>jb|Vp+h%rFbmqET4
z3RwlHrSM-m{=a9sH7KmG&7sChCoPN-Op%n6|L5@AgJ0?VJ~JI<4GU$CxZWaQ#2U%V
zx*?YfLl{Z^Kc@pfz#MUu33#ieu4dY?(5>5m${Z;1fszfTc(Kqy0Y#jL>%iq?dQO)#
z^6m`$3R&pqN?&1f!voGg3W<NZ*yixdjUK;P|4`WxaCq1L4;psbs-cnZiVlkUe*t{q
B0^k4u

diff --git a/backend/main.py b/backend/main.py
index 02d793b..7bd917a 100644
--- a/backend/main.py
+++ b/backend/main.py
@@ -10,8 +10,39 @@ from pypdf import PdfReader
 import pytesseract
 from PIL import Image
 from pdf2image import convert_from_bytes
+import pdfplumber
 from sqlalchemy.orm import Session
 
+
+
+# 2. OCR Module
+def extract_text_from_pdf(file_bytes: bytes) -> str:
+    # Try pdfplumber first for layout preservation
+    try:
+        with pdfplumber.open(io.BytesIO(file_bytes)) as pdf:
+            text = ""
+            for page in pdf.pages:
+                page_text = page.extract_text(layout=True)
+                if page_text:
+                    text += page_text + "\n"
+            if text.strip():
+                return text.strip()
+    except Exception as e:
+        print(f"pdfplumber error: {e}")
+
+    # Fallback to pypdf
+    try:
+        reader = PdfReader(io.BytesIO(file_bytes))
+        text = ""
+        for page in reader.pages:
+            page_text = page.extract_text()
+            if page_text:
+                text += page_text + "\n"
+        return text.strip()
+    except Exception as e:
+        print(f"Error reading PDF: {e}")
+        return ""
+
 # Internal modules
 from database import get_db, Email
 from scheduler import start_scheduler, stop_scheduler
@@ -57,19 +88,7 @@ class LoginResponse(BaseModel):
 class NERResponse(BaseModel):
     text: str
 
-# 2. OCR Module
-def extract_text_from_pdf(file_bytes: bytes) -> str:
-    try:
-        reader = PdfReader(io.BytesIO(file_bytes))
-        text = ""
-        for page in reader.pages:
-            page_text = page.extract_text()
-            if page_text:
-                text += page_text + "\n"
-        return text.strip()
-    except Exception as e:
-        print(f"Error reading PDF: {e}")
-        return ""
+
 
 def extract_text_from_image(file_bytes: bytes) -> str:
     try:
diff --git a/backend/requirements.txt b/backend/requirements.txt
index 9f0da77..8c7160e 100644
--- a/backend/requirements.txt
+++ b/backend/requirements.txt
@@ -9,3 +9,4 @@ psycopg2-binary
 imap-tools
 apscheduler
 python-dotenv
+pdfplumber
diff --git a/frontend/src/app/ocr/ocr.component.ts b/frontend/src/app/ocr/ocr.component.ts
index 192a56d..c743cd9 100644
--- a/frontend/src/app/ocr/ocr.component.ts
+++ b/frontend/src/app/ocr/ocr.component.ts
@@ -51,7 +51,7 @@ import { ToastModule } from 'primeng/toast';
                       [(ngModel)]="extractedText" 
                       readonly 
                       class="w-full" 
-                      style="min-height: 300px; width: 100%; border-color: #d1d5db;">
+                      style="min-height: 300px; width: 100%; border-color: #d1d5db; font-family: monospace;">
             </textarea>
         </div>
         <p-toast></p-toast>