From 8c47d6d9ba9ece3445bf34c88f09d776f3ea3dca Mon Sep 17 00:00:00 2001
From: phiresky <phireskyde+git@gmail.com>
Date: Mon, 26 Dec 2022 23:26:20 +0100
Subject: [PATCH] adds test for fixes #106

---
 exampledir/test/twoblankpages.pdf | Bin 0 -> 4643 bytes
 src/adapters/custom.rs            |   8 +-------
 src/adapters/postproc.rs          |  28 ++++++++++++++++++++++++++++
 src/test_utils.rs                 |  15 ++++++++++++++-
 4 files changed, 43 insertions(+), 8 deletions(-)
 create mode 100644 exampledir/test/twoblankpages.pdf

diff --git a/exampledir/test/twoblankpages.pdf b/exampledir/test/twoblankpages.pdf
new file mode 100644
index 0000000000000000000000000000000000000000..464779490715eb9a7b39f5e43c4a998f7f9d92cb
GIT binary patch
literal 4643
zcmcgv-EP}P7Vb3xy~9BP0|^j`9R7+$5Fobe)Y!OnWuq;s=3+#SB*qj+C8;Rct9`@0
z*f-dHk^Rn)ltnsiY;3Ux#?%bYoZs*K@Z|dZVk{=Ub@K23{PSN{Ko04}d$P~fhb&QV
zO8LoBMaoB(-@db6zO?K!o$K;uD-%V0XJ*+KSzfD>=Jsq?yj8`jUZ1N(r|O+$H(T;7
z`{yjJ*CZ_aG^?v?Ri5e1Ru`(MX=>Tuscf~b@%Gg#tEx*SH`Z5Q{e$H<J36Qm9jo%E
zJinJ4MfR0+j@a8byG8wJi}TBkT&YK@eq`~nckElk>6afC?^RNNS@rOVs$XuP`puQB
z?uf01qYrNSpY3QP_e2Y(tp#%%*dhB?eXLtgp1nInC_Xl`jlUZmx#xFFfW)EZf3mR<
z42RCO^`rH<h<nHU&Pu&K)gNij41zeIsqX@2*Ro_JF-N%RcD>syh(U|ZOjUYUCaNNX
z-ZomzhUIcpW~`lkUFzgU0gs21E0t#QGvXSzT^j_Q>7s`0ir8+$sLLABX?_fv!%G+B
zCb<SVWp4eSlS*{SZ(Q`KtU@CHb!a?x&{MNY^or9Xbidp{_a?23-G6Cbb<ulbVu`Lc
z{Ap&1Uh#WuYKcEiEx`$M55mXimSFULF}Vbp;PkW0nA#jYDZPyGz5H@}+Fu|Ko09*X
z%0H+&OC-WvtIba|MTdD^C0rpQf!`OoTvg<`7I4=bk6kB18LDNTC9jKBt_Y>~Dyyoj
zST&Hso+X@`YA_aINMqLxNH~r|u_sW2nvgRQb1k!NtxI%Exoz`=MWf7g1;FhJz}$_d
z<OKl*L1@-=8dDfX<buTWyk_AnnEGUBamG(x-JLHz5A@){e9VG8;UEWC7_SHdd6_-1
z5kUa&jUd>`=PYOn1q?0Tn|m$Y;a<xl_Wgzz^9R3934Zz^3<57U@}N9qbia+v)*(z|
z=<uGOEWe}cp6sZA--EX~!*AA+&${J#aA4@bp~eCCEz9L@a>#>ISt$-9p~4-I^j22&
z*;<w)+z`0x#t05N%89{91?8Hw=!1X^P?0V72;sq_Y#Dl94;Ee1Fr~Q7@yHL4`sWGV
z$PbF@wW4HKn#Q*(<#wGVRMoOpBkOk>StIL&&P%x`NsC-2ca&#^qDrkc$b%QOV`v*5
zsanciUN`UeJf%%WmI&ZVtWERN!_RzV-SpVjka(xc7nI6st<s(Vu<&V5P!$c;NH(0Q
z*67PyouOfltU2KE4jxNeI_5)LS!0+i8p>UPVUcw40&Tkey<GOR^Eu`R?dBajbbSaO
zGL%&5JZm%s7o%W9*;i2*VLf9@jCj&ihJA)PR+1RALR7-=g)UR}!!pZrYz;wf2HHiw
zQ|!$N0M@SA8(2F7*;1E?SE=t1Mjok0mNP*go<yiBpe<LRdwbww$XP%~?80!*fSYim
zMAT1v3IiGd$+1cR$tRLUtDtNc9n^=KjeJzqiA%FLypL!#vRDc)V1Xa(@#Dp3$;etX
z(&0|jy#}2<RB}M~YgrSVwwD#*7bAF~R#{On7WEPbS;aoV{6DgO9DQ~ibbI~2O%T-M
zUV^-)nrG8Rt^xd-rB-Xbujp%Jbgt=VUFPZ6oFU)q#tp=R74o9ejTUWIRIGNBKUqQd
zpGM8Ciob&3_}h_1WM81d`E(0>688nOZd2YpI}-bxZr_irZ*k=_UpJyRJol&Cezy$K
zg1tCI-x&<JrgE9$!TNAvGfk5R+@}5v#@4yZ6=%k(<MHme7WMk;-?Fr#c|&XFh8Y=c
zaRr}*JCsX&<lRfCjv2Y#cw_Wp1**fWZoZ+;BwDXt_3reBoiktpxT0s<bXnh#J8@hZ
z(|*67n1e~8H_y!kpY9TTiayIY{^KnQn;r}F*58^F5KpS*q)_!Z*L%6j6yN;qIqqKR
z^c=;Lo}GW=I<9bh;W@G6I`OY~X~bu`_(O{O#{RKXOA9q=YGE{J|0oOsFQ6s0XI?1q
zk->jdbZ2HapmV}00yw1e5rp9xPqKqE4;^-RCZ_061G*r5$S;IuLptn-c8I`xC}SD}
zNdp<v@PU59W#IMcgaa=P&wK{)fX)vd<v-7lfV}=)*L&zI5xEcS5HZ7haMuqW*yg!_
z*MQF48pIOeKtA>PdC_q!>r!U9DlL;Fm&H<Jm@(;V&$O;Fxip)b?C*-<WnZ4t{3U>u
zMxpRyC-$P{GL8cg2oWjas6+(5|LP?sylA;txRFdlm89W9IZ5n$OFvOz;-~Q|lk=Dy
T%GzMpi}AI7a`NpD7uJ6P1{mI;

literal 0
HcmV?d00001

diff --git a/src/adapters/custom.rs b/src/adapters/custom.rs
index 54bdca8..ec6cd7e 100644
--- a/src/adapters/custom.rs
+++ b/src/adapters/custom.rs
@@ -21,7 +21,6 @@ use tokio::io::AsyncReadExt;
 use tokio::process::Child;
 use tokio::process::Command;
 
-
 use tokio_util::io::StreamReader;
 // mostly the same as AdapterMeta + SpawningFileAdapter
 #[derive(Debug, Deserialize, Serialize, JsonSchema, Default, PartialEq, Clone)]
@@ -307,12 +306,7 @@ mod test {
 
     #[tokio::test]
     async fn poppler() -> Result<()> {
-        let adapter = BUILTIN_SPAWNING_ADAPTERS
-            .iter()
-            .find(|e| e.name == "poppler")
-            .expect("no poppler adapter");
-
-        let adapter = adapter.to_adapter();
+        let adapter = poppler_adapter();
 
         let filepath = test_data_dir().join("short.pdf");
 
diff --git a/src/adapters/postproc.rs b/src/adapters/postproc.rs
index c178ea0..d318ec3 100644
--- a/src/adapters/postproc.rs
+++ b/src/adapters/postproc.rs
@@ -221,9 +221,13 @@ pub fn postproc_pagebreaks(input: impl AsyncRead + Send) -> impl AsyncRead + Sen
 
 #[cfg(test)]
 mod tests {
+    use crate::preproc::loop_adapt;
+    use crate::test_utils::*;
+
     use super::*;
     use anyhow::Result;
     use pretty_assertions::assert_eq;
+    use tokio::fs::File;
     use tokio::pin;
     use tokio_test::io::Builder;
     use tokio_test::io::Mock;
@@ -243,6 +247,30 @@ mod tests {
         );
     }
 
+    #[tokio::test]
+    async fn test_pdf_twoblank() -> Result<()> {
+        let adapter = poppler_adapter();
+        let fname = test_data_dir().join("twoblankpages.pdf");
+        let rd = File::open(&fname).await?;
+        let (a, d) = simple_adapt_info(&fname, Box::pin(rd));
+        let res = loop_adapt(&adapter, d, a)?;
+
+        let buf = adapted_to_vec(res).await?;
+
+        assert_eq!(
+            String::from_utf8(buf)?,
+            "PREFIX:Page 1: 
+PREFIX:Page 2: 
+PREFIX:Page 3: HelloWorld
+PREFIX:Page 3: 
+PREFIX:Page 3: 
+PREFIX:Page 4: 
+",
+        );
+
+        Ok(())
+    }
+
     #[tokio::test]
     async fn test_postproc_prefix() {
         let mut output: Vec<u8> = Vec::new();
diff --git a/src/test_utils.rs b/src/test_utils.rs
index 8f45278..a981fc2 100644
--- a/src/test_utils.rs
+++ b/src/test_utils.rs
@@ -1,6 +1,9 @@
 use crate::{
     adapted_iter::AdaptedFilesIterBox,
-    adapters::{AdaptInfo, ReadBox},
+    adapters::{
+        custom::{CustomSpawningFileAdapter, BUILTIN_SPAWNING_ADAPTERS},
+        AdaptInfo, ReadBox,
+    },
     config::RgaConfig,
     matching::{FastFileMatcher, FileMatcher},
     recurse::concat_read_streams,
@@ -41,3 +44,13 @@ pub async fn adapted_to_vec(adapted: AdaptedFilesIterBox) -> Result<Vec<u8>> {
     res.read_to_end(&mut buf).await?;
     Ok(buf)
 }
+
+pub fn poppler_adapter() -> CustomSpawningFileAdapter {
+    let adapter = BUILTIN_SPAWNING_ADAPTERS
+        .iter()
+        .find(|e| e.name == "poppler")
+        .expect("no poppler adapter");
+
+    let adapter = adapter.to_adapter();
+    adapter
+}