From bec85f38f4d64a8f14ed8b2ed83e6278950f18a2 Mon Sep 17 00:00:00 2001
From: Roberto Rosario <roberto.rosario.gonzalez@gmail.com>
Date: Sat, 8 Aug 2015 04:49:08 -0400
Subject: [PATCH] Text parsers and OCR backends are now used in tandem for each
 document.

---
 .../hybrid_text_and_image.pdf                 | Bin 0 -> 15612 bytes
 docs/releases/2.0.rst                         |   1 +
 mayan/apps/documents/test_models.py           |   4 ++
 mayan/apps/ocr/classes.py                     |  40 +++++++++++++++--
 mayan/apps/ocr/parsers.py                     |  24 +++++++++-
 mayan/apps/ocr/runtime.py                     |   5 ---
 mayan/apps/ocr/tasks.py                       |   5 +--
 mayan/apps/ocr/test_models.py                 |  13 ++++--
 mayan/apps/ocr/test_parsers.py                |  41 +++++++++++++++++-
 9 files changed, 115 insertions(+), 18 deletions(-)
 create mode 100644 contrib/sample_documents/hybrid_text_and_image.pdf
 delete mode 100644 mayan/apps/ocr/runtime.py

diff --git a/contrib/sample_documents/hybrid_text_and_image.pdf b/contrib/sample_documents/hybrid_text_and_image.pdf
new file mode 100644
index 0000000000000000000000000000000000000000..cad612ae8169e006752469d654de1fef86660449
GIT binary patch
literal 15612
zcmdUWbyOQ$*EbbPixzhpC{`>42?Td7?ykWhxD;!VQrt^%io3hFxD<CSUfiMh7kck~
zt~~dXzrMBJVTC!fXP<qxo>`flAC<g_7$b;@4VkJfv#qhMx-Au%1;7k|8CoFo@Bm*!
zZA=}_0IYC|B2diS$`NW06tgmLgo;3oV8&2BK4b?+d#HgmvTG7!U<2|)5RS;z6|NFW
z;v8}G&J!VIJKP<g2PkC@^Rm<GKd(BG9<e4t3LgVCTN!mxFZJbVxQcA=T0*!+vm=fz
z1b||QG4v<0`V76hBpOkHR#76lImK1bbDh+B>NjM3w@BpEQ$H2a+3<Q}$^^o%GTN^F
zn8ppS1hP47%K#h5P#fdlm4T1$^+&!}`+fjoWxl`ri^le+amsGCP@uY;p#{{)5gDlD
zWaxNL5QEuUBLjtmV6IvKMh*yo6~YbxLs;(5Odx;`GEm$e=41=^jS^7V-oVDe*1#TW
zW8?-DRsxDZoz0D)isC{*X@H}>)4hBC@E{92^1t{G`L7P7S7w%He!QIS$dvP(Y4Ulo
z029Ki&c~1c#V+<gZ4ed&Ff*|u0{{S^k-@JaT;w0Na{S5ndmuUCl>b{Gc|!!z1(<O}
z&Urfu8NIIfHRrL*aIlb`mkITLJyoEnNaTQ_f@E>^_tfz?fm1Z>l<F$M2tdkG!yXE3
zal$E_)mzCA)?}vvIN=dqUu$XS8OYY&gHty1_dbDM4BZT@BjTS>kYN6cS@+8Q3Kuim
ze@uY<X~JK_^`B@-S{s-`fokT)@GNELU<OJ+%}vc50U$PZppdzvgFMt;7-nq?vw_+;
z0yzH&2IxM+75^<DKw+2_%wEaXzzF(hH2)T*-FY_w8S@uN2Z?@WJA~9kERo6ORMUZ-
z&EHTvgUdppRz`(AAKCk=0zz~?<mP^e5mYw*Ho_Dnb>++ftKsV7J)d)zbAkmor}9J2
z&*mJw29Mnb98xwQdMytT{-+(ei3fdsmPo4oKcQ1^bw7GOLBF$J=jWc!K+Mvids{9=
zusKxl5nU<ymWDq_y1@)#v2e#WM`JhtfS?koo+{3{2ceh2?y@T)U?EO!XSUI?5L3gN
z>qQ`DEc{NhOgd-WBC%9~277Q72Zf>o3qN9N@;K@=awg?;*@L*BeK}|qW|rz|HCW@Z
z!$n5=bDz6fK}z_=({BA8=NDbCCC%fiPBO1#i+!zB?Arsqhzk=Zznf25^7kx>ioFuk
zoqi=L3XrwEV{A!}pYmVDxz1oVY#}UH5v94le&KB|&U?A*9IJU5N<_?7n4=chW$bmS
z)KQe*tx84r?Ht3C?nraCsaN499@a#O`|kVvK-h(8eNR*x!%?b#Bofy#-AgSM?RfHl
z_tB^FEt4e8+bbA^l^dY1s|QE}+#49?aJBrcRLVsZkYvnnGV_H?a+H5YDL?k9nVrQ;
z+T5dAF$;~Pz5UW7toCw9DK~<$$7P)et43*^n2eG>1q=vkIm~%bt`gs^8g*jYH8Hbu
z8Vu~4RwVWBe>Oc5_!^V_eae-(1%K^M%<3#n7`IWL$`CDD-Fuq{t3A7IjsOe&xF!Rk
zHSwH_*ZHAVDo@H|InU<O@_BUoLdl`E(RtZI$o%uuyrv<sr+(gHNI5ZEIQ`{HQ81nL
zHn&jZtxpS6iFsK%B(&I%_jdw`ZFC!wKhs62*u-sAR`$XD-8(HsP|Xcf&VOxMV6PxS
z8yz)YNQXaGme8^wOTW7;5Fhczp-|T#VIhgbgMMWKdmbyIND8C{<<(dSQJcUPtfQsO
z1zcib5v1sjROw*3lyhTDBO7OF*HBL`jHkF_KWfj9^{SRSjaN+lot>8Y63fH6L;j^T
zS&9Gd@lMpu?DD4!ort2h<SPrPGlg^LmXAh*CPT5ZJUwoHOpH5-pW7@}E;yg&k6Z3m
z@AKQO4Ft<g``;PbnY<o4RdTN>)TyvRSsY#yROafdcxb2TnAR2lXm+ARqrz+Zwql_K
zO9joew>uEyu)+T5d+|3*>xsg}>buMaZ;^x}8Z7^Bs>>%o0KlBK25?c8hINA-@m)9K
zewRX?sGNN|hG|j$A#&W)F`Xx&IEuUNg!_+<m%BQhSHju`T-p?+w>6@(De7FuWuk62
z$8{!^kX%;ZcjS*(6IUOFCpk48Gf-3AN(!bdORBBUAZF5Du!Q=3WVYvreD-&*jxzT4
zayrjHIWO$H;kBHAExoWi^1w+^%zYwwM(n_c=$<my)N>ovEBS$HSY5?&xwgWyGS9K2
zVAvtn^mXJ6o}|x}NQMQEe||zp%^gsXI!F{G&c%um&4-1Z(|q`Skl0F5X-}F+YxZK@
zJAf9_Zgdv~(jiuh<u{uO3y>`T9OXJ3Wj7>84|6>-J+UZcJhvk&X};(e4#&b(;oO^+
zdcM(@I#I;8y_q|1o_@1C^>xRYS{BW21xPYomhx4Jp@KQ<DYV<Gy7$AWz$;(G2k|4m
z<zzs1<BR>7=)Fw*x`;x1A#iofm1E-WzS%|yA4`N>S1<c!0qthi_5eRp)sT^w=B%qv
zTgBJ&F6Fq)Rf>3dzb&yVTh+|2kCHmXw70JB=-@>}_gZBO4YnCC@szX0VS4PdPU3{T
zYOX9&Izfwc6z*D;XONpP;)@}6vhts<L-=*k<fx1DDRyR624rq4(orGIz}rT0M#tkj
z2!|1M+)t_}kmW(9mn}v~U#NZYNSjFqCOG8JM9aDw*7R&w-YKaa%IO`WxLz)6wC*|(
zcd^s7)FsR4bKMC0WPT7Rad5{wCogeodVpONGNt%bFlW|6{E857ksd<HQU5+QQs(8?
zOlEJ6E+->mQ21E>=f(D1(Z2Q2Z4kv~fS#;hS+=#!v{cfyt<sC-c2#qBlj)QS8;-`N
z;-o3?Y?M{6CE9XJ_v#xT))yt4=Ei1bQzEGoLfc-8?GXo~Z+C%EF)L)fCkk)+N3}LC
z9A2Ayr;mrs<J<6iri$Il+nLO5BTa?_Ya&#kMXIdldWY_0N(YPnz)Y;D9lxR0MEvu+
z@%U?&X{%F}7^0i0?fx)`PLK1I`*z?>0xe4DWlLQ<Zv93{JtY-hPrE3slv5wSX(|tX
z6-{rJHIL!+JL#Gid~RoKKT^6ji$>kYvTmI-;urLA*dA_S>}wOA<2#vud_m1WG?OWH
zN<=*JDa0vCgB$RnnXQ4n%vVq)dI_b5)S>yK79Iyp09pkSnCcvy%WB&vynaOTC9YJQ
zEXQSFw^*74>a`_oI5e!GW7(sNxOt-^sb{E9{*<4mlxtSQN*`>oB)cIFnyc2J%jL7K
zX@Z{RihyzoKbQ^eZT{5V>wPNZ5`m5F=Jf5ojD~*2Zfjuh-h#``qepx0-1}NABGnH@
zEG;IyPsR`}?ZfTQYFAzmta^8y=6?H>TA3@H!01k~=r8BIEB6FeKsE4g_yiNhPAq2e
z%O<IrWWmMqojTV??ebXW5j*DLio3(`J`hWe$jQa!>!WivEytFK<eUDth)>#cV$sTR
zGf=<gdLu>8YZb{Rrm)IoQ8Knx3-3)ALrxYYCeYa?h3q2_^}!nRgb)h7GY8^w{|Ot&
zR*93lZ0%ev{xz5jd@F6J?V2`Q{K`6z(4;QJItBL-6w(_NqfUumSO4rQ#u>Sgc22+z
z>-w!ix_X@CDzDm2PfDQGm%N?>rk`kIp&=uY0*7&0$K8i-RZ)t&Si~pn7xa=Xg-GO!
zIPTc{vU0lQLc!KY*eX@~p)!m}>Ka5Wz4G>P1nPm<H&*>lZb4=oo4nSe{;hJoiAmR6
zqh3hTq_>xAz53~Lnr>;*6vr)_4~EPZgg@?m;{4*k<)k%cYzLU__{M8S$6otoLTsd5
z1=n(-=qY91JK_d;#>BuPkG5<stw4Sq5?ST-PL#dANp`#IHB1rk(4HGxy*w|yO<>w5
z=yOk7B2g#c><r&P+T8}t<eg-RScO{+7Vb%tG@}+_k)gEcvd9~=p|fNq2T3W0s9??c
z9}VJ!46g0Y8!tn61O{w}%57sXY`?vCAq$(9idw$fz-YJ2p4F?+QxW{ZhmtO3=>p%@
zK=BK*1%b|{`;D@IW;#~0y2!6qS<}*}$ZPGc#V;n5y`%*a5=@kwuRN4pRS^$?*N@sX
za%hb&2dRL14LbJH5h_RmHl)b&!(j}8YZtB)=vvMPK_=~>#T$ERrfwBe`pvk;xuKwI
zX(#WN4xZQ0YAio2?xa_!!jjh=58A=1jW*(|Fpk*oF;9c^o($*EW9OP;QPY!Qj()0G
z+}n<ut;#bjTgh5RHz4d&4H+WX*QwMnvFKK{2wJ63<+y|n6Fc~Nj$5CN#;)#4RT)fv
z;yyCxY&^E@Q%ye)``FHzv?q_>QDC9C|N5naqc5RLyj~8QS9>Z`c)n5%^R$$896<~g
zHD6t=n^=p9Juv38@_uu_@Sd|4?i(ueCKa0<A8CZ{LR3}<{rwV2tbudNFoW4rf2Hau
zo2TNuaz)2iq+%p_y{cb7c&>V6zfs!n<>!p(ebaa%Ue1Nbq+Cb&uquYU=87Y9>m^vv
z9;gyEn|K4ubJQ2T<1o!4#)z%6I%Qva2pJYLEp;FMXpu{$l}8m7tEfF6rGpd|u7P^z
z)xR(s5zjJWr<ZB<fSq)~UR3^5k;pK*PM~k)o0K4>qa@zbiO2StX?u4bg#N0y{Dy+n
zZ?BKa?E>FX_FR77uXGw1Ix{X{D7fH(iM|xr8H+3$*;}=~jOjW_^BMMErT6?`<QSAx
z)NI19CBvHs*1Zy{E>ejtiYh^)O1((5dh?ABC`C{oz2%8(WBHsZP(TJ6L3z=iMU8OF
zc{d@Qb2wu9!&;S3LPdAEJPtnV>}3j$a<x5h25(`jM|JCiaaN2meU0EM^Le;*opAxW
z`&Z~yb^<!Aj)JydRkC<uRoKLzrs4ZlI)(x_7DtT%!;2Lhem}VSf+?Jg{atozwPE%Z
zti-SD4*hu_BI*g<lyH;ITei40?dC(J?{q1u^o88*_0vd<3rYgxE+dW7+mlqIf~=@q
z8;aQ7QYulo9JdFuwYoaTatqQI@VU_g1w5e<nG1JUG+Xu3-0SRx=oNAw8H~vx!qtj%
zF}UUR(4T|kId1eMp1odDjK;YvQ|yvw7#3XcjOtN^04Io|y)&Awj}Zy;IK-k^dmNvz
zE^zo*oiavdwm4}_tDT7z^ff=7K6p;U4{7f1mGBaMAnewyYR-q<)6%i+%%i$4rYKt7
z>v4iVE$fhZuu-}k@pMe{{YmS%@>P2Q<Z^@t{d~MzmB^+yNOhi4Z%!<t$hiXdi&c&k
zj?)Jc{S_zPcJ<lJ!-E0fqX4VeJ&st-ahfYtOgF-FQ-#s)_)YBTt;A?-HxnLfsx;NX
z=t?nd0*56R`vsHv6o(77>&4D2<6z{ymfmTJ?~@^S*-G3VpU7g0O6kgD`%N1-z2vWY
z`Qa}52JQARnG8&VX}RxosE>zYl*dxPtRom_y!rCb-l}H3+u_A0Z=I+e?o^(%y~pQ|
z#dgeiemUsGRJc(Wf6aubuNzrY=o>w(;id9=5<tt-xBZ#TN}k5IO5SH8AQ~i0?C+<A
z8mix8YOG?N6W`GGZC~zpeVx-9)?(8Q;Kj69)by9GR<tCf+!WmA8c_`>Ww#flr8MnU
zEsxZ0+@h8|tl52yj&rNubh*>yau`n$$Vu~xuG?v2?5rK9Kx+5nWjBAygsUt5waECi
zRCd%E=at501!#s?Vl^24y+QXWYgDvi1?O3WVq<4Su-K{T9KE{l%ct{_Hu@v33J86t
z^ss5VY0sw+#`2)kDk0vQ^zh5$gd&gpvl{^j<wK{Bf+XIa^(Xt|FyT;yO*Gz9T{l*L
ztE@5!CTFKBx*k^tuQDo)CwNb8@KZfI>I>CYmXZUct9wzk$38Ix#`_+bADu|ibL5ar
z=O;;=66GZlV~}ioT3?Wh&Uic)lb%_iDxxP7a3K%YgY>DguqK3jVuvRUmN$<W+SR0d
zWp)+;C*{M_Si$FBBJ~CSvKUwRJuv5U08NZa4HsS(gC4n-Is9p$GoLBmI5k{hvDcGO
z{>7%lCtis<>;<o3d#+t{_Fa|5>~O(tA#pk0kgMy)QdH58tI7>s=PWlRtbGw{oEiP~
z8`Sta(P}!HqjD7+_oE;s_ozJ%|0%K81sj__8wu&>=qBwNz^U1oyd-D=PKn(r7|Pzo
zs+W;&@x;BeDSnK?scyQwql*eNHwQJU@-6$WL*mc+l_Wk1C8dD|Ufv#1ro|10IPF3e
z&a{+Lr1td|^`(b({Yi1=<sg3T>rVzlj&QZue3Ne(@;hmP&~Kj32S?fMpJT5eAThMa
zqTax3D;2l5tZ|d{>y7DW={BZv%BpW*(k^~9vl>9XX}T4@aq7)bXH(g6wb&xcKsB#y
zaF<Q}j-3JcZ1t@)g}gYSNa}o!RM%xF{uLE=j==N!5`2i5X4UySdC?2xUpmcgeLS^S
zD`^}uV>{=&8qr;2BVBN--GxTlh1Zk34^j=6eAadzL7zac(Y>o2WThgzYD%x`OV~^Y
z+oY1LF?}A)o_}%^SHGI4o6)UNwv$|R6jMYAuY>MWs0<!ozdl)-{RWj}I{X{2e}b&s
zEpgQuf5k)mXdS6K1F5Q5a+)#0`bj&LzIk$%qPKvM8Qj<$qc7flDIYj-qh1=*$4fBD
zI$?LY_tO#dm9T!H&y(0MAGL)$tyWI|?6_(*tm?_}3tFjWjlqcs`U;D%I-<eG&m!=E
zBA4Tas}+=dJI!V2SV}zI`$=H<(0$EjR&SUokJ%77?XWaX=pyL+^%Es)@A-k%5iTN>
zM=LPe13yYG@|32|^WzJ;rBWpKv%YVnZZYpO&XGWiD)Np#F0-@G7YoO4<THJ`#5-R3
zQtl30Y({fW+{8X!Lci%AsB-4GdR{&Att<k*A@qpY+vOAUj;?k3UHq)02U{(CEp^(i
zCcXCUN`^faD{WCx$bc8kHrY;8rdGv=cI}=qrcZSljVBQ&G|tI^qY7D*x1|)F7!RiN
zPa>x2<Zg<it-|L$4&!f^6(m)yhC0wzw6HLysx${%I$Z@Lm@my>q+41RF7XG0>+|jx
zXT)m#7g=fx*}hH*5xZK1J*o^Cx&DFS;e&eud|&pgt^o0j?hbESRL*97eWAL<*_MX9
zeIM`kFX;Cs=X8Ln{C2h*@1YJf;X6hXMg|5*#>OTqTaKTcx{nnOqTgQek>Q7@)=Ij&
z3oG;i<_&e;U|Syz@}6-VK6@FC3dqQgn*NB-V8pDA((|A^)`}88qh1ng&ReB4nW#ZC
z*=u7l&O0xIx=kzSm{>To$4pvod2&mG!O1IDNb5<$b4U9WLN_L_r1h`$byVdlkjUe+
zA20CUA2m#=mY5DTI?aL^6UMTd2tMS%>c00&R|ghN@1vvTQJqt{RVIWk*a3S3G+lIa
zZX)vNo%LLQYS4N{I~x8F4v*e0>~U#T*)6xda|7N@mTk6E;bnEb|LM-P8$TL0*Dvbb
zercrUk>W>#Wv)UY5$5GJxm$T-Y=M?Ge8M|W<D2ngzb?CFQA(4I_lBCk%3?Cl!R2@_
zn;I>zLcVIAt!pD>@zsFT2iG^}38PEjMm3{@!?*4l4F{j$?~9YaP<$LRO3{PL%U=>a
zw{1zdl`!ZM#<WVWnpJ{dZs^^xbzWhd&5c`*y1WA%2EM2*+@_KdeRH#`z1Q?Xwm0%d
z2TM@VxOC7T?U8wCY>ySv`Be#Dc5slggulq99n9B5Tkd!xF*4M_x*#?nsd(O#C^DjD
zODF<|?7N+Zirx{&)~T-7N&ez7!B#PVqx0)SSkfZ9|IFc#+2ulCGmy{3wu$_^d=ysb
zYPu@NCVSZVZ9_Erl_@UhcDs5`i;x{>VL1e2&<3(w6+aRhl>GiChjG4@-)oxtyNnf=
zbQ=)<NH7T_bZRUXq#xZ|pXVal9-S%9epz$b(Z_EzAnO|EK5lgxJzO|xI{oE%a{ozy
z=RkxOr%rh&E-gUn*|`gEwP(l9Vj&|p5%Jh~fmGA;3MuPH(R&lochjC1Wg{HgYaX>#
zmSWz~mx%0wHG+L)85=McuP8thLi%|?d`Vz6Cq6I;G83C4A?WgQ+@Ut=+-(E+^SLM~
zo06+U)l$mpONeH5{@jR#nO+4R|5&kDHrZ~%6YDpjTD5(r&>(GAYSax@y&%Sd&QFO9
za<}c9pE*cWWy<zu(wAIT&DvcqWw=ksSlxFq!{Y))Lu4f{3r2$jDqnF<bFmZGZ1b9g
z&F8s;7>w$)`y$qPQ!cQ~^VSQ@lVubtedo5DCCA14227b=ldnmWuH<rpGxKjfc&R#!
za?^_fK76GOMq6<LLSEl}SB%Px3R_KA=2$M2W@$^>P~c@^N?<}WYzUNhaYb2}u2IoM
zCmXyrX0UuAoJ6+n={kP*#KopqF3ib7Ws-x1l+&PRk6ungngVrkRIn$toKtu)R84z1
zj(`pfH=aBGVgcc85P;^gS7>yOa?yB;O7vy<@=j)Vz{YJ^rcH9f@@_Q(RnW&GNAWkG
z@gL#JOb`crc6+4J+4o5KbfP^Kh*l6L%P)}kz5W=hwTAYk8T8hthwjn9tHlZ)3~SFU
zAE$1FH1cj$`FB}w?i5Sq^G9zb&fF|mMm)S|sGlcr(|x)|Mj(1;g@_wRxo3Rt^JAmj
zt)BMm>jximj`1hS2FaIb2njy$+XSc79a8gWbaQO<;Xf({iH?0PK90ZHA8vKVMMy&}
z$5l4=`H-<nRQq07l^fB+n29>VxP$kctK9yzr6IL*QM2=TG~02nD1tt@RH)nD_OjVI
z{b?9=o8H7hxXwxAIS<Rb?d>LnUWnw3Lznrk`Rt@}@#s{ojJOi(XxQ%S1W#baOpQVb
zKk4<_c^Lu%npBhEB3WqO?Tb5uW^84Ws1S)3rNp~SC;bh0CZz*QCN5-K%(;V8X@Jwt
zy&F2nM>zju8V0t9-Z&BFLI2Y9`PGZL@A`n*z^wmi`+%7Lz3=lss!az1{b|Jiu}=s3
zOH=aqMjhzB$M<iYy5D<PAOHl!%JSFt;TptJUTZRbBdGu%>&H8?r+!1R{;ujmU$OK{
zmng`vzUpf=Y8&832s~Ux(wF=`Wo|Jt<@UYSZnQ9(ZNWOB$9WXoP{%15P&xgnzjofa
z+IlXza^5OxqGEe@PxfnPB3HbuQk&O?)KsVBO#JS^PIyz)z{SF%OY?qXZDw?ZT}<#g
zrmu%lKssyU5e|#rQ26t&7e)@|&PC&04?PIU4GB`?IhB_@kNGhTy@Df@dF&2;{G3><
zMF(Ue#PXaP^AMqsx@SBg&mem06fN+yh%ELY+2cmXBl%18_81@D6~R_|blGsx!L<yJ
z))x=vY-!(g^>a|th`!}V;pE2U#aYDlMs^kZAU?s%nk>~Nzeyyd`f_cMa)tH9jwjA?
z)r7Ezd78n`AsT#vXC2R8Vo7I~lbqb0V5W+_!NK7m$4$VTLBlv^pv%0OX&{~*->X`@
zfr*j#Ifj2O$-o~nKY2pb%kq%c56cXXwuY}HJ@sKMA9*FdCpM!;(6s3=)&Z+M=E?In
z*Gj8GKNL=QE*sEvg?>D*$2N4sY<gFd_TsB@yx6`GVLh=YMqx8XWpb)<5bb9|{MJLQ
zTv$3y_p3@@?FXxNz2r9OV)(-l9@LMc%mbqwVsk%18RwrH(sqTT&CvCiFgb2+Q{@L{
zVEW1~pGkk9<RX$bimpJfRm9Lwx5f4RlbY#<=C)s(M@-8#%XFUtA01J#J!qkQUlu+3
z8pQ(N@&M)O#kKF!yW8f4l%YKx?uG_u$0ytmZ_lbH@bp+G$7(&bgl{J5XBfwA)M6{r
z8~b%FW{s-|oQSHV;&UQ4x-+rLK9CbYZmI?R*7_(KbNlg7RTbv&5QR?gg#1k3;Y`uC
zRbUE+%+qc@>N%XH^dyvAM3XEuAyzPojXs@KzwO!uJ3rgsefG$sejeO?ME|f+>8&XA
zVHDDbab%*xMb7DgBhIgv7p-%vKKvu-XLz_rU@a|e-8GS5d&IZMZ+{}s_EQvpFGYwz
zzN;4XF?Jhc9GtotCU~*+>}pm<bz48SB3f-`1`*E!`}I*2SeNdTF*ci{V~~y`6OgAi
zQ!E2h+@>Ed*bCgxsxYLZe~p52en^C*XKuMcAFc!X+P%^^I<dDwI%c7^kH)XVp<}3P
zr8Ya|kMSI}tNLh9Oo~l9a%%hq0Qm`^2?gZ=F|qmw29dalb#d#$nd;I{i4Dvv0jL@B
z1eG5rnjL!Ek;q6^$0H>h8M&tfVE84w)A;Lwysffg?;b9iBvvgLkE0g%`SFE@0x%=c
zyYt4Ski0B<0>mCHZMc6WOcKq=KJMOdMm?ewg7j^Z+#sy=mpItuid)HlD{ek`2rPH$
zwDRgc%%kP!M?N;^XY`l(M$gZtVU2P@RQ-jQw!8efiYddFqh%He6tnq!1XllihFnCE
z3xjMe&^uoi8NONlN33IEi|gUWEe_FQ%&n@Bi?o+Ym$%IOmPoDm-zK_S)OVkRnmv|^
zF~jfeYOrQ)9GG~Mk;2B{oepHkH6C}>Ar7Z7(qLTJh2^AZ+zDm!Vn3{cZa-`BYqX5R
zw^ScJ)>Za>GMiWW8Esu9-kY3+c!LpB2~~B5lNhYwwg|v+9l{I47zunrnBR!HSLh?2
zQh6HCaA~E(>)v)cEx>|mA!~fTXwr!XhW59{tU#R!iK(ego<ANJ(0_z5rAhpGEz6nk
zdMv{Bom7s6<xApD>PjXwgM-*Y{)Lp-jGs?347rkLe6}YjaS*w!1*4sY65R5W9#2_K
zfXuqh6n(cDI?=fw3_bmXDZv&e#<ekB%4kDeDO27E#K|_w*u*SPl6fC-Y*lp0p<hFZ
zGU8X*sw-~Ba`Cp!(bE0`{XxsSCom(I;)-d+_@V;XJE9sW8pVgkXmLm6t1WhFgVWJd
zD-;m}%(fuBiQ}*7gW{7%lseTY)dN#LMZh^)i8gfmxJ`o7*zkH=HE+vUY;eVdC~<R(
z>2SdQ{w%SVB$`_O)m7WZGivL|i7y?B-~IIrmkOWf`YFDOq+k%u&C(Sx_G#RtdLeZD
zP>u5WV-KrmFW*sX<ruJ4i?Fbra)lzu#C*PZOyrp+)hXEx+FA4hi)|m*ed+P<1|+X^
zBM`m=u9t=S?o8E+zNV~eoWp~y;a00k!Lm=*QFVki6+aP9y)8)U!W;9NMe3wm`XRPE
z)ZYHc_R&X*NeZ#O&SS+$tBwj<%$9N26mo=g(f3guF&?^fJr|xUZm&0WLsg7wEDCr1
zbQn9ri)4e4yv~c9hr=<O0{KeMm>gv0LaQ^pPkFN{Vl*unLWF`zUAd`63HmI9EV`d@
zG*9}U7Hi_0Plt$#B}aDCRjH?h<04ZH#CD_Er_+_iJa>|m+*@Fj$$d7dXH!2TB0ri%
ztvJ8x8<sa!fD~|P(Iod`B-r@!5P44T-M-y2gl5mBGpT*A{RxeiK&VS@c%EeOj0QPf
zRM?n-c1~3&4mnD-dXPx&^fsAdi9msF573N2)4SB&^~1<}bee#Gu@>#ahs{|;XlQK#
zUWkM<W=)7xA(L&9GemKj)$d>GhJO|?qIWTvrKTF*mZE%<95n^5leoykORR3%uNc^i
z9~j9l8%({vR#!5*Bu45QaH*O~I{r+G^e{1@Gc1*dZ(=Osa$93B?8d!_^DuqbU|jWy
zcnxBJR=)PBdl<$p7oWOLrp8Tp&h)pP2`2Hr6zUxJt7dl!VN(pY)vcf-mQyq#Qgqgm
z^l!Q7yjvAyv~AfB#UCgTaX#uX#m!HHw9uuluCMT*byD;*6zQDBJFp4%>uZ!X77Hu#
z>xGMNm6eb^D$NpBYp$%`spyXVa&$D$RY*@iyZRItKZ6Svt__Xwvp-1`k_b&u`@W)c
zXht+ge}QJR^PYjrkBxd9R^-Z~l^+m5!Un7Sl7f{!iSwgqP3pbL(-9hr&IpO_I1zvA
zDv}AQ*bv=~(=qD~{!<zMd=oXbDms|0vDTG*Q(0!QgMQ{^<%niW-J7xIyymsB!Tssx
zoaUdO&p;T*`F6AFlc&-CDsB15iFOW?A}?#g={dSffAl5Zq%2e1?1aZ&j$>gC*9(!r
zKwsya0F<wcn*)gA_b+9GEDmF5EUQGc3reacS<;C-veq@xFeiUl?y!fXle8$^DnRPw
zZ#5yWOrdUqqKIBO%NfY_pO>qU6$M|Vvv_3fXcA(^ZT9UzL%PFCrb?y*r;2N4cSF(<
zT9ghH9@NPnXg+xLoPpqg`+Yd-s?QSpBanQMqxmippMN9e2P&dKm9M#R6Su=IlG*4i
zL<~m8S){SCr=945XN4cTtF+%Hu1HV?IFWp`{aK^Nd5uBh^{$~iGKcf}X+TTs5?g>c
zWKIqdRh!W4!@|aa=$_Laa#zyE!6-@b%9-to{tpHkrnS;{Z4~HwD+hyhdJd};0fY97
z>)p@L2kl%V+Ss+Z#d|rEfFEwYw}muJLe{r^+T0;YtF|F+<mff_wK2YHk(h`*-s=b3
zB)5D4kf{z&#KB~YGy4n@^BnTGxzEWlT`lGAR1RM4@rc{7*U?x4v6qb4nlT{j=3!4-
zlhzKN(Xr4>(XVWA3e|dqwF-D869O;5xNd$g9UC1;D)tV9iN<lmJoF?wHOO_?JeZYa
zcsFYm!{QqgVx@e!moprRP#&!>`USJr^hXl7__!w+3B}XSLd2R<dZgJ9t{Kr&F@5F+
zNWVW#1j%e67GUm4h6#jvRK_riEH;bl%d6E|CYhz+0U3W5E|p4_K6BDtXI5|^w|Sqq
z;VN6W{9TR<*7cfK?6kI-d@Of_NSeNB`BJ{_vn;SeGuZujta~MI3wIQETj7p*)U}Bo
zF=ZWBy)r<Jr?e*&6^(R&NAePNaf%efJA~)xSqK^(`n$c_*gA#pE4>=Um#0ggaqClC
z*Fg^(kcjN{r+Cpv_0}6O+en1Z^=C|K;%15x5?Ja<rO|4ZGzuFhf*y5nyI)D)RG2u!
z95V}<=B&8uxeoLv1tJ$03n!Cwt$Ep*<>xA!z;`EGDKEuKKL+MXOINa^U91%rXF7RL
zo>us)sQ6RPanMF@1a4{CEMvtBIW|QYqH3ulx>q<=B`z7HmZe)bcH+y`hSvwShsc?c
z4L=Wjk}A)W4hVVJM@g_@g-plGhar~YhGC8XSdBkSI@CVWKK;OoP%-OE`b5DFsXV`^
z@H9$A1cihSR41g_wthsW5Ju!rfZMy%=jeSl)P{CuI2d7DPqpNlF>7Cqqbbq(=EyI)
zps>*oWJIpOS-~#NPDN0?+J&x9ILuMGM~lMGxm6T@ee8E$ia&sHeUiWbG|<|Da6#8I
z*Xp?Tyq&b))nI(Wqk=E4;kcu2O#1vWPLOtR6{qR=4`<?dofXqF>xtYFOJfLZgj<A)
zMBn1jJjQqLSmgsZ$y5603)L8;0xsz^5#4@gpZC*S#3_P2QEZrlf{_c~u54bc3q+hX
zFJJC{VMTr&AyoAARf_UK5Y&Ga6+sO*1fZ^xjM@K932A9ZNP@)h#PL<@jtl2|^l(a2
zbk0)0Ym)anQ?P;P&5|NKz6(n9$7X^Z#y5PpKF44L_Nvyv#4{D%w6<{bG9h;<Y1C{M
z>>dyv5q)Slv1oaDn&{_J&PZ&B;{lpvwd9a!v9aox67G6mOw@+Le9wtXANcP$KO?47
zmVU$bWXaecJGOWLH@wY`pqPsPuocb!qik(xtB&tbaJTBsrMaefuvzCPU6De0E`|~K
zuNtH@wd&G+KI$yxPnUEg@z)JG4VtxWWt{R>`TsNsefeRrd%w{oU6LVMdP{{`IA#BM
z5pub7h1=#0nhYMSJD;1L_uL>`%0dyVuX2%${b8g0fp5hUaLOVhPx)mq{-FFBN1}7$
z2kq%EevX)=!<(5O6j&MBx(O6)XuIMvP}e1J>6)^Uh9lh)oA~XF*R#}W!@rEZw#g&2
z|G*JDZedmRZK~jW%bF<p-NIvi4_7tg)q#fPZ!JGN&Me$>M600g1xygZW^I+*iqum*
z{)G{`xXNxPIydS855G^|Mk%7y*Fmt+q`kth4>B||#@@qa=_NRzWeNt8mal=_1sxZ*
zxNpb=(3-H;9smZlx?e)-)IUbAb<^e~q#P*we-4!lrj#nHl*k`V=gHz!OSOC`LH|H~
zb8<OavrtF#%4N|P8vBLRuo2K1GWE2S?A%9ItyniMu>gNXSAJ7I$Vn<oy(Qx*KJI3#
z4H*Gma0EIggeB2)SY6(DB3?4Du5->NEMD$YDp-%L#irSC;PZ(hKyd^`pwY0|<Rgc?
zMJUADOC(LoGR=*YfPyvI*~Ai_fJ7>S%*OAP+ILv7cXP4mV-;fx2A~dv8pW!5?X<#C
zJkhe0K>30Q_1@L@1P}&-rPhP59fX%wr|9nAncrSWs>zOgz8yY@5RYCQ67X(4S;%po
z+o*L-RrpbA<hM5$Fn-PmH1#TnYE;iFqR|g8z3o9m84)ZAHlwXT#f|?)i_~U6CM^?>
z&MKaxdOmmBxiw$l?eKK6SU0(Iv2nXY+GS<Ix1hy(TulY{dvZbWP{E<mPw7%;nSy;I
z_LIwT7{$^JWTNpn3&wLgkEybP)E(vKay7Flqt|$GX;-DCNSh5~gO)C^V*HRSD)2gq
zCl>lyU6W*g#AJb=HT0<894Cei5b%_o-drH|Z9p#JTPFcuNKn!bI$eET%v~(rmkW<m
zAw4%5!S-V@r;^{JcC+`rMUhK8pN<(eZoFUY4kcig5G!YX*K%Z1&g}d8hXfwpyZ$x}
z^^lk!c?=AKbnT`Fuejbb85<CeEWKw)@{Msd#ZwhPy|TH+|LQ-vyYbolI!;49kTd<!
zzPiXAu<Z4l3T82e=v4n_iYzG%{b^jCasHwJ);Rx;+)t21f2v$RNUDF<4<d(GD1Gw&
zKMe5(1Uu4QDoP)Y2SYVD@jc-`s^G`CQS@L!p&#C$?3&3o4Wq`!jyA70EQ~r;R)ay^
zS3qC_8Sd$UJP`h@{Q-dfY37tM53bxmhY?eCL&f;w?8suBJAR@>nF2UDq;okSz30)+
z$xpGHy>>GmLK&LuoT+A47SIB5>L*?Tnd;2WWu-FLs3JpHTrKG$wp=!Cd522VxQaux
zd&9wtXOGiUB{D}2DW792?Yu1*?0BEP`BOO|RJLR4mN=-Ps|(x6?j0wdTy_S4CppDW
z{(O)cJ5Sv(<c)NlO-PzG7lbt#?G<Tia0Pe9rXvqKFMGV~pp`>II0<Aq$i%-(V1ga&
z*?kk>x|zXxsLm7n>61@n1NCSCy0fr=(yON)ss~Sp=Mt(~PjGNH^u;=RvX|{XCdj?J
zIdnx8N|cn{4-e1M8SQy?9@+lRkJ|_cVqa=Df3Mkdc`;h4K%DTN-4%yQ;h{{LWQ-U?
zzJQVKtU)A_jnB9ly+9LfnBfqT0uE&of>Zjd2>S;jcf|J3FB9<;AL0e@q2DB>NmDfl
z7T{id(L~uoa^;vo9P<Ob@Yf^C%m5IJ2c*e;{TAf^qw_B9&SvNZ?zOz^!_}O(vo!Kp
zY0~U8LkuDkjMZ;TKxwa^=|%xGsM=2htgk`0WDlg~zgxYlHCxcmUB`C1`?SW-_%eP@
z?y2gC&u)mZP$@40mZy)h#876P1j<78<GfqBgIZ#qFkilJb22@kFU~{<A#3JuX(}}4
zF$m?5G20(Eot9;_9(P#^Z0KW|ES*?uqz`|4#^-{3Qpub{3sY=083J=U&Dq)vg`*gQ
zHOcyBoSqGlM_i|FWf}PQKXnU1aY{j$e*I0y=bJEdOML}l0@gb3x1k+IqyE+P(REFM
zAFT8n-=A<-mRIJ~Zx82`ABLlNv1lZ|b0#_>!!O^(%EkUlfGywlSfHj;oE{vRHBa{K
z4Fhr4vG-<Tk2b|b;rB!Al#ZlsdUa$P=ZYcSR*FSHvj=I5Gik--;bFM#kC!}mxAEsa
z-a18<JT9kSRm!iLBF|IY!}Gi_3KrT}D${kJ6^W>tsHSK~cB||!YoDM`T=%WL=__f3
zVO)<~O(47>=6Ju5;+^Q7_<3@q3D$@6t_H+<fM-;i8#leSwHCj2u_oOY7gxg6x2^{{
zM!uMho5vgP`xpoRxzqgM3-6Rb9$%F~^%U~YG!$|XtQ)0UNoz@0B*(E`Iv=~Zr5M6f
z7;Cio+)o)kuUihTn?HQ%{XU%kdZ^8Kec{4=aIUXEmB%59hr`7nEK?}<<L0XXXueTJ
z){rGh#GCjB&niqlC+udN1qm3pn-V_w?CD;>=*~^Ys0q78nA<@T0puDoyk@OPj5i@O
zWMWp-faSK2JZkCtTtJ>aV?DjQz}V;z>a6i)C@h}znETU!Ko-@M0!+b4ookM?#OXSJ
zr3t-90_m2{<Ix}@(_;d?v=Pz+M5)4e%W3AV-^Vd>oAMv{3?rPSKd7;g5mnB|9Jj;0
z@qH2L_6To<%EP3rWpL?-W>oR1^I`cAAKBvI?3LN}_H>{0-o|k2g6EQGNDHUy(TVw(
z#ojWnrvaZK+0L>N!cSUejJGq5W&%aNdCCd8-$-o*k!<laS3e<+*wP7$ggJrjGi~Yo
z6<_V7@jQ5pECr#BSIk9!@@)wx9c_*4@hjvk#m=R>C){buGXM5w{tuW5I~W4~^UXZ?
z&zIW25E)`H8%GhSgOR<tts@MM4uW6G8dyVtLW1`P{cCeWsJ(%sIm|{0YHw}=2VGbh
zm^uL1-~%}4FPIG@hyx5@WMyUrfS8%t0T51*4p7q3z{=c6(8kmX3SdSC3OX45LU2Gp
z$Z%Z7Zx@U#ATTme*ueI8%m;iPyglFuwN?di!l&Geha;AtEO0Q${k!yg6vrRH7|?$s
zV_3kf{{qnY>kD?Bh;J4hfam__eExm29iq`!807NybrGW8W;004T$27`9iK!Onz<IL
zhy0IZ)=5}R)kk%;G9I~~syaXakrsNyJli3i#v3vdk)A2lXIpacG1}Ho6jn3vAbSF}
zI{&GAa1-QbtPNT0?$Wzb{E}8n){$@GqaF2HNmoA4UZ*9Ev8yn*yQKKJ<6$#9T5I9X
zauv>+U0^qspqm9cMT$qNe%bMR4D5fJx!jmNzN?DOyrs7vMS26^_Yd!-po8%5Y;{u|
zl}R-AdB`-1^BuMaTsKL2^m|Bn2r(%*3|x9HORuvI*xy+u&=u{{FvbR-gI?|7z09_q
zY2Cj}cIxrpAkK7pi}Y`K@&|4Qj*0qDM*Ih$qYTGE{h|sPI6&`X^1n|Eb9)CzVKW2!
z`+&bT_<amP{teyZpv4Mi2e7k)0c>on01gfg06Q}?fCYZOzh+_oZG2CIL$vPK`7<X0
zp!=MK<9(E2DmLc#xHZ791o(SO`~_(w1_FS7A*KGBGc12;N!`l_fbM6&rT&f}gIh0Y
z4FUWj!jWO{4%V+GmLh!22KbeM@MHptx;lz0Il=*SaCT8~C3uVISC>ff{sL~^y~KMq
zd3%_V64X%(2uJ4um7%VVaM^I6lkjgR$=^;wI)BygFU_W3{+7Q7M#<lU)8MN92H;bK
zI>4Olji3&IU&bf`WuV692KPv}d)vV9RvZU}Md$ws;3oMyzzq(VvT=kj0T=nZ?tj$S
zKlh3M-}>B__`NUx(C3#A?i-DNr0Kmv4#>cNItBSpufQez$1V41C2el(par;hv?AbF
zO5GEMVNP%l-?!raSc&DIp%6BJ?`AMlcr@TE{vHfupqwq#M$qVfXVm&lBv4&L6JX<H
zW%b)XX1MUz1~#U&P#Z=SrB{EAHOpThn!<4WC4A%M5fc>?gKwW4%*-4t%*<d2{LBVF
z!z+S=<+qG7b~1wg!N(^28xI)H`->mW#|EDd2LJwn&j+)@KX5vng_#vDi$(PQ443t5
zNx0r3@I0m!;bLJ1u`_eR|3DxP5SWpfjfR<-=G9*u;%W~yL53G7D>7Wt?*{<P&d$mX
zFai8V1F^Ai!o6^R0BnAz!N>PC_757%eeV8~#`fPdHdc6u{>BGlV`G6=*xzVytKkoy
zf6~D0|E&v{<G*MetoPf*-{o<z-?wi6L1TgZa}78^5&U1aaolfjf1e9tWo7;68gLr;
zpZ0=S+1cS4|F?BPtelYl;$vgx{4aTIp!<~l`&>3K$3N`=v2k#~H<Ca29PJI@k5~Kq
zdQvif1BHhds0f3>O93wHeqWTdF@XVoW$OKf2tW(KBPuM&$pYeJ7lE(H$p#S-=71w>
zSs)x>F%b|13>M)-{=c)_*PIB<NZ1T&Wa;2!4FC#*AtLb9;}m4)5CV&WSw&bxIoUY`
kLBb-e!e9;-q5Hyda5S)Y{FR0vHV7LhG8L7moEY-|0@;bZF#rGn

literal 0
HcmV?d00001

diff --git a/docs/releases/2.0.rst b/docs/releases/2.0.rst
index a2dbab348e..4adf77e92d 100644
--- a/docs/releases/2.0.rst
+++ b/docs/releases/2.0.rst
@@ -77,6 +77,7 @@ What's new in Mayan EDMS v2.0
 * Support to share an index as a FUSE filesystem.
 * Preview images' titles are clickable.
 * Improved API
+* Text parsers and OCR backend are used in tandem.
 
 Upgrading from a previous version
 =================================
diff --git a/mayan/apps/documents/test_models.py b/mayan/apps/documents/test_models.py
index eee2553c99..b5839b4f3f 100644
--- a/mayan/apps/documents/test_models.py
+++ b/mayan/apps/documents/test_models.py
@@ -48,6 +48,10 @@ TEST_COMPRESSED_DOCUMENT_PATH = os.path.join(
 )
 TEST_DOCUMENT_DESCRIPTION = 'test description'
 TEST_DOCUMENT_TYPE = 'test_document_type'
+TEST_HYBRID_DOCUMENT = 'hybrid_text_and_image.pdf'
+TEST_HYBRID_DOCUMENT_PATH = os.path.join(
+    'contrib', 'sample_documents', TEST_HYBRID_DOCUMENT
+)
 
 
 class DocumentTestCase(TestCase):
diff --git a/mayan/apps/ocr/classes.py b/mayan/apps/ocr/classes.py
index 0f5a4471e7..13a5eaf499 100644
--- a/mayan/apps/ocr/classes.py
+++ b/mayan/apps/ocr/classes.py
@@ -2,13 +2,47 @@ from __future__ import unicode_literals
 
 import logging
 
+from django.utils.module_loading import import_string
+
 from converter import converter_class
 
+from .exceptions import NoMIMETypeMatch, ParserError
 from .models import DocumentPageContent
+from .parsers import Parser
+from .settings import setting_ocr_backend
 
 logger = logging.getLogger(__name__)
 
 
+class TextExtractor(object):
+    @classmethod
+    def perform_ocr(cls, document_page):
+        ocr_backend_class = import_string(setting_ocr_backend.value)
+        backend = ocr_backend_class()
+        backend.process_document_page(document_page)
+
+    @classmethod
+    def process_document_page(cls, document_page):
+        """
+        Extract text for a document version's page. Try parsing the page and if
+        no there are not parsers for the MIME type or the parser return nothing
+        fallback to doing and OCR of the page.
+        """
+
+        try:
+            Parser.parse_document_page(document_page=document_page)
+        except (NoMIMETypeMatch, ParserError):
+            cls.perform_ocr(document_page=document_page)
+        else:
+            if not document_page.ocr_content.content:
+                cls.perform_ocr(document_page=document_page)
+
+    @classmethod
+    def process_document_version(cls, document_version):
+        for document_page in document_version.pages.all():
+            cls.process_document_page(document_page=document_page)
+
+
 class OCRBackendBase(object):
     def process_document_version(self, document_version):
         logger.info('Starting OCR for document version: %s', document_version)
@@ -17,9 +51,9 @@ class OCRBackendBase(object):
         language = document_version.document.language
 
         for document_page in document_version.pages.all():
-            self.process_document_page(document_page=document_page, language=language)
+            self.process_document_page(document_page=document_page)
 
-    def process_document_page(self, document_page, language=None):
+    def process_document_page(self, document_page):
             logger.info(
                 'Processing page: %d of document version: %s',
                 document_page.page_number, document_page.document_version
@@ -32,7 +66,7 @@ class OCRBackendBase(object):
                     document_page=document_page
                 )
                 document_page_content.content = self.execute(
-                    file_object=image, language=language
+                    file_object=image, language=document_page.document.language
                 )
                 document_page_content.save()
             finally:
diff --git a/mayan/apps/ocr/parsers.py b/mayan/apps/ocr/parsers.py
index c8285499ef..deab577b7d 100644
--- a/mayan/apps/ocr/parsers.py
+++ b/mayan/apps/ocr/parsers.py
@@ -38,7 +38,7 @@ class Parser(object):
                 ).append(parser_class)
 
     @classmethod
-    def process_document_version(cls, document_version):
+    def parse_document_version(cls, document_version):
         try:
             for parser_class in cls._registry[document_version.mimetype]:
                 try:
@@ -56,6 +56,24 @@ class Parser(object):
         except KeyError:
             raise NoMIMETypeMatch
 
+    @classmethod
+    def parse_document_page(cls, document_page):
+        try:
+            for parser_class in cls._registry[document_page.document_version.mimetype]:
+                try:
+                    parser = parser_class()
+                    parser.process_document_page(document_page)
+                except ParserError:
+                    # If parser raises error, try next parser in the list
+                    pass
+                else:
+                    # If parser was successfull there is no need to try
+                    # others in the list for this mimetype
+                    return
+            raise NoMIMETypeMatch('Parser MIME type list exhausted')
+        except KeyError:
+            raise NoMIMETypeMatch
+
     def process_document_version(self, document_version):
         logger.info('Starting parsing for document version: %s', document_version)
         logger.debug('document version: %d', document_version.pk)
@@ -139,10 +157,14 @@ class PopplerParser(Parser):
             raise ParserError
 
         output = proc.stdout.read()
+
         if output == b'\x0c':
             logger.debug('Parser didn\'t return any output')
             return ''
 
+        if output[-3:] == b'\x0a\x0a\x0c':
+            return output[:-3]
+
         return output
 
 
diff --git a/mayan/apps/ocr/runtime.py b/mayan/apps/ocr/runtime.py
deleted file mode 100644
index afe052ad29..0000000000
--- a/mayan/apps/ocr/runtime.py
+++ /dev/null
@@ -1,5 +0,0 @@
-from django.utils.module_loading import import_string
-
-from .settings import setting_ocr_backend
-
-ocr_backend_class = import_string(setting_ocr_backend.value)
diff --git a/mayan/apps/ocr/tasks.py b/mayan/apps/ocr/tasks.py
index 631833422a..0980faab79 100644
--- a/mayan/apps/ocr/tasks.py
+++ b/mayan/apps/ocr/tasks.py
@@ -11,7 +11,7 @@ from documents.models import DocumentVersion
 from lock_manager import Lock, LockError
 from mayan.celery import app
 
-from .runtime import ocr_backend_class
+from .classes import TextExtractor
 from .literals import DO_OCR_RETRY_DELAY, LOCK_EXPIRE
 from .models import DocumentVersionOCRError
 from .signals import post_document_version_ocr
@@ -35,8 +35,7 @@ def task_do_ocr(self, document_version_pk):
                 'Starting document OCR for document version: %s',
                 document_version
             )
-            backend = ocr_backend_class()
-            backend.process_document_version(document_version)
+            TextExtractor.process_document_version(document_version)
         except OperationalError as exception:
             logger.warning(
                 'OCR error for document version: %s; %s. Retrying.',
diff --git a/mayan/apps/ocr/test_models.py b/mayan/apps/ocr/test_models.py
index def2d78150..14525af663 100644
--- a/mayan/apps/ocr/test_models.py
+++ b/mayan/apps/ocr/test_models.py
@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+
 from __future__ import unicode_literals
 
 from django.core.files.base import File
@@ -19,7 +20,7 @@ class DocumentOCRTestCase(TestCase):
 
         with open(TEST_SMALL_DOCUMENT_PATH) as file_object:
             self.document = self.document_type.new_document(
-                file_object=File(file_object), label='small document'
+                file_object=File(file_object),
             )
 
     def tearDown(self):
@@ -27,8 +28,10 @@ class DocumentOCRTestCase(TestCase):
         self.document_type.delete()
 
     def test_ocr_language_backends_end(self):
+        content = self.document.pages.first().ocr_content.content
+
         self.assertTrue(
-            'Mayan EDMS Documentation' in self.document.pages.first().ocr_content.content
+            'Mayan EDMS Documentation' in content
         )
 
 
@@ -56,9 +59,11 @@ class GermanOCRSupportTestCase(TestCase):
         self.document_type.delete()
 
     def test_ocr_language_backends_end(self):
+        content = self.document.pages.first().ocr_content.content
+
         self.assertTrue(
-            'Repository für elektronische Dokumente.' in self.document.pages.first().ocr_content.content
+            'Repository für elektronische Dokumente.' in content
         )
         self.assertTrue(
-            'Es bietet einen elektronischen Tresor oder' in self.document.pages.first().ocr_content.content
+            'Es bietet einen elektronischen Tresor oder' in content
         )
diff --git a/mayan/apps/ocr/test_parsers.py b/mayan/apps/ocr/test_parsers.py
index 3dbd97164b..813ac339dd 100644
--- a/mayan/apps/ocr/test_parsers.py
+++ b/mayan/apps/ocr/test_parsers.py
@@ -1,12 +1,14 @@
-# -*- coding: utf-8 -*-
 from __future__ import unicode_literals
 
 from django.core.files.base import File
 from django.test import TestCase
 
 from documents.models import DocumentType
-from documents.test_models import TEST_DOCUMENT_PATH, TEST_DOCUMENT_TYPE
+from documents.test_models import (
+    TEST_DOCUMENT_PATH, TEST_DOCUMENT_TYPE, TEST_HYBRID_DOCUMENT_PATH
+)
 
+from .classes import TextExtractor
 from .parsers import PDFMinerParser, PopplerParser
 
 
@@ -46,3 +48,38 @@ class ParserTestCase(TestCase):
         self.assertTrue(
             'Mayan EDMS Documentation' in self.document.pages.first().ocr_content.content
         )
+
+
+class TextExtractorTestCase(TestCase):
+    def setUp(self):
+        self.document_type = DocumentType.objects.create(
+            label=TEST_DOCUMENT_TYPE
+        )
+
+        ocr_settings = self.document_type.ocr_settings
+        ocr_settings.auto_ocr = False
+        ocr_settings.save()
+
+        with open(TEST_HYBRID_DOCUMENT_PATH) as file_object:
+            self.document = self.document_type.new_document(
+                file_object=File(file_object)
+            )
+
+    def tearDown(self):
+        self.document.delete()
+        self.document_type.delete()
+
+    def test_text_extractor(self):
+        TextExtractor.process_document_version(
+            document_version=self.document.latest_version
+        )
+
+        self.assertEqual(
+            self.document.latest_version.pages.first().ocr_content.content,
+            'Sample text',
+        )
+
+        self.assertEqual(
+            self.document.latest_version.pages.last().ocr_content.content,
+            'Sample text in image form',
+        )