From 81018f06fa15517917c4c6e52d0212ca669b35dc Mon Sep 17 00:00:00 2001
From: Aaryan Kumar Sinha <sinhaaaryankumar@outlook.com>
Date: Sat, 13 Dec 2025 01:30:33 +0530
Subject: [PATCH] Added --overwrite flag to zimit

---
 CHANGELOG.md                     |   3 ++
 offliner-definition.json         |   6 +++
 src/zimit/zimit.py               |   6 ++-
 tests/conftest.py                |  14 ++++++
 tests/data/example-response.warc | Bin 0 -> 2272 bytes
 tests/test_overwrite.py          |  83 +++++++++++++++++++++++++++++++
 6 files changed, 110 insertions(+), 2 deletions(-)
 create mode 100644 tests/conftest.py
 create mode 100644 tests/data/example-response.warc
 create mode 100644 tests/test_overwrite.py

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 58fb40a..2a99b30 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -7,6 +7,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Added
+- Added `--overwrite` flag to overwrite existing ZIM file if it exists (#399)
+
 ### Changed
 - Fix issues preventing interrupted crawls from being resumed. (#499)
   - Ensure build directory is used explicitly instead of a randomized subdirectory when passed, and pre-create it if it does not exist.
diff --git a/offliner-definition.json b/offliner-definition.json
index 4000466..4bb68b5 100644
--- a/offliner-definition.json
+++ b/offliner-definition.json
@@ -970,6 +970,12 @@
       "alias": "name",
       "pattern": "^([a-z0-9\\-\\.]+_)([a-z\\-]+_)([a-z0-9\\-\\.]+)$",
       "relaxedPattern": "^[A-Za-z0-9._-]+$"
+    },
+    "overwrite": {
+      "type": "boolean",
+      "required": false,
+      "title": "Overwrite",
+      "description": "Whether to overwrite existing ZIM file if it exists"
     }
   }
 }
diff --git a/src/zimit/zimit.py b/src/zimit/zimit.py
index e982cbd..b205007 100755
--- a/src/zimit/zimit.py
+++ b/src/zimit/zimit.py
@@ -849,6 +849,9 @@ def run(raw_args):
         warc2zim_args.append("--lang")
         warc2zim_args.append(known_args.zim_lang)
 
+    if known_args.overwrite:
+        warc2zim_args.append("--overwrite")
+
     logger.info("----------")
     logger.info("Testing warc2zim args")
     logger.info("Running: warc2zim " + " ".join(warc2zim_args))
@@ -1036,7 +1039,6 @@ def run(raw_args):
             warc_files.append(Path(extract_path))
 
     else:
-
         logger.info(f"Running browsertrix-crawler crawl: {cmd_line}")
         crawl = subprocess.run(crawler_args, check=False)
         if (
@@ -1091,7 +1093,7 @@ def run(raw_args):
     logger.info("----------")
     logger.info(
         f"Processing WARC files in/at "
-        f'{" ".join(str(warc_file) for warc_file in warc_files)}'
+        f"{' '.join(str(warc_file) for warc_file in warc_files)}"
     )
     warc2zim_args.extend(str(warc_file) for warc_file in warc_files)
 
diff --git a/tests/conftest.py b/tests/conftest.py
new file mode 100644
index 0000000..d51650d
--- /dev/null
+++ b/tests/conftest.py
@@ -0,0 +1,14 @@
+import pytest
+
+from zimit import zimit as app
+
+"""
+ cleanup disabled because atexit hooks run at the very end of the Python process
+ shutdown. By the time cleanup() is called, the logging module has already closed its
+ file streams.
+"""
+
+
+@pytest.fixture(autouse=True)
+def disable_zimit_cleanup(monkeypatch):
+    monkeypatch.setattr(app, "cleanup", lambda: None)
diff --git a/tests/data/example-response.warc b/tests/data/example-response.warc
new file mode 100644
index 0000000000000000000000000000000000000000..143b947d121e61b479cf9cae596b3853a9a60633
GIT binary patch
literal 2272
zcmb`I`9Bkk1IJ13mB}8VF(i@kY%^vP#gc21YjZ|(tK<xG6RQ@vm%@}Iw;F4;hmeN8
z%YB8GklZF$j>qG9zOV21Klpw>KYxDuyxy<(2SO1MX-n+EA32QhCSDKm^z*;tt%-oa
zbzumE4h)IVMxfD1Jx$oZqp70_gTvu^a6R;OZ!|(H;-5YrE|U`=pae?&C9rW^Ya*p1
zbjT~mR!ua~n4kol@A;&6KjUFCyxAaDN7gUAWNv-;u#?O2K?`&&1yHTSX2yZ~do0BF
z1v%l+=Di-I;tyLt)05F<6`<!YUJr+kSJv5y>6pGKI@`J(9hMq1{(0}2gk8{zb>Pdn
ztf2h@%$|g4W_!Wub6PgVBlKrY&?+_I1R=Rwg_rSHm)Ep2@w(m?{$*hGDov_TiqXL}
z0tcpoDm71%@65h^YyJA;--K40^~0$xbQB6a&<_M!xeX7W*;F6p>U2!VxOy1^y$a5Q
z-pjO~<X5GH2s)7@MV}<)T1o6U5;}v-BWQ=3L%EO*ZE>%c9tE?}km)g1?1MA5sn`*w
z$lKUUV=a7qe20eud{<SR7*u{fKFI$%^-bFwI6Ge%Asz?=;Pm0RSAcVZ5(dAv!TmO`
zEQD(qIUIW=WG<$0cwi*-iklIt<LGYz`T4B3`LAHF(ZN`=ri>_YdRm1V!X&#}%_2<;
zXY$Dd(Rrsz$A=p4{*@O+`xQl->mGC&Kt>|U@w4i`4>?aKf|ZRw&fGDFxi((>swq|x
zEJ-enG&wo9g*lq8WCo_E$T*v2AT1<;SSM-eXF!jHexp|6`4&aoOOY-yjl?Oy14%U_
z7x;tIX6YVO19|aFr*itiluYb0eR&EgH!r|eIE(wiWj}%a%kT15Qt`cPy}M_!gXME^
z-*5=?N(cWyD!?to@xoeU!EuGI<4jAb=T8aHF|S$QzZ%#@x?aXqtJfMaNU0)v-@=d4
zVhUJS?IT!ujK6MK%=`r3gMMQ{Q`-za=tu1&>2F7R1@bzSQU+6QK5he86a*^%nVmee
zZlYH*D|*?IPYec9Bf?!h)a3;k$h*MgqrHm>&Sm&Ams+1QAlFv_EKas}NY1s%CFspf
zFi|5sG6(Z*1$gY>LEu?3NB|;sGsrO61_Rh<Y9Ka0K9FZ$VJw;};{S-yGVbdlI_~I!
zRY2lx{ZP6JrwT9YRSd9D^v~-j0@BW=_^7Id(>kBY$gXypsfgb@9Hoh>$adRqN~4Qg
zl_)0UBp=D<QEB7sdm;QtcIKv3m@a74C=(0tAM1h&1Ceji`+D)!A&HLfl}JV8BH#UL
zhu!%ZE?R|Kf!x2^RnkPr;<D91Cd~KAhbT`WPvg~h%QPUoKBndjF|4&}GM=A;j{E&K
zRc1j=hU6QwJ?pBn)lpGtaQ3GTfknMFnHfp0JjHswJ?EK*;+3k(*W}}Mr*EQ0<d}y_
zf}hIJeUPS>sOvUiy`T2mO49{tc%+KMV`<X^ztZLG@o5c0d7b@M;baMKE4s*<o5|Rs
zDDm>7?C|dMl|0gd$8u(Qm&XJ+++rn_k~<pkiOiIT@#M(YZu@BED>PGgMojC`zD!i@
z6PGKwA7;NuQ<DZz8tlw;yAWK++ps0>2=@C$oRj}X)uO2txn)@yLoe1YAGD@@j;}z)
zb<{ILaFg#+BRTs7{n=Ng7}lyrUF>w~f#n2oN$klv((};%fk97HV_I}_!G=+MwVqac
zXinZrw95%0>+%D}#`Mwxwx#h`z5a+FBG<JRGCZ5D`}Lq<hP2A%NtyP{65ztp&LnY{
zf#Y$+be9`4fns^yL!|}j0d4~!QOrBP#^BwtsZMWnM@_@N<3zn^5Zms1>%+0A{ff1h
zKR>m03#_$;o-QX4m(;D+HryntdXKkob^StIv-Z|`I^t2mx5lfc79i2WC#rOFYs{dI
zg|<)8ih=r}UOoqvsUxlv)w0U;xHvmDIzI6k{LKK0bd4?1y_;Q9i%t|4<Z?Q3g?hEx
zcMfV>)6CQtB6g<Vk_n$fT-Hd|ZM&1K{eeeziQO02RxJ#5?BPmt{sG7NGVd+sucvir
zVj9Vv*q&u+Gpfqoubik$sAH8kGg-=p_aTc<KKt+V)j^hfB(`sJwa-UQil3@}r$E~s
z3#Qn7{iDEnN{dqPvZ3vQo~p8P*!6gU|5%XiG$r65_<yiqSQhun7Z@;wcY0PnMihBv
z`OnTA3(&cX3C@~MiKJw(Zx+L>U?})Q-i!JMs0gDfL|e|fCnnTq)tcbTzjc4<%Y+EE
z#ni*By<S$ep||!TfYm{m1H=|Fh#SxwLItKtFOL0qUaVJ#W=BbhgXbDg$o4keMY(Su
znfv0{pV0@(29nLqiPD#$(71Quvn;{n(Q1uU-HwM)5f9G(D*<;v?8|!0shr0QA@3k`
zjF)-)J!2b5*A$zi8K8)km>(?Y{UA8f=x(Iv?)b53Q5t8Yr?andDyRP*7EgFyQH42e
zuacpvQqzv4Y={{<e-Bm=uy*S_@z`y@&@u+Bul^=8SxkCyCN3Z?79GXp)&3=9e?u3*
zHebfnh5oF)#Lt3qAR|$Wm>0&aenFiM`WqwXlHV`sOAP(2mztg{DvB`LyC!M@VAyM7
zS14cud+eaoBLn0xaRwvtMs`zsRA<QY!_~Z<oU0d(9Rw*wI$ZB+9dTTLw_%u0cSF!#
z$NTOF3fn}|nGf5v?Mos+;AUl359vtSN-~|rY^N+^6jTh2gqf~p-l_hf59?}9$<swQ
z4~`osS}3T;mXha(wiBZ(>dHj3)HX^$@~-ZAOq9ancd24=pmn<vW@7#sS+OUUIA`Wb
z8vt>Crrvma*d4A#_O6AbcZTEp5`+$D98#-XBzbn{k>%2>;U*#I@X_s6KHew>eu=87
RH3|9e>Fz|Z3l0B4?{AQ;Bs>5B

literal 0
HcmV?d00001

diff --git a/tests/test_overwrite.py b/tests/test_overwrite.py
new file mode 100644
index 0000000..e41baca
--- /dev/null
+++ b/tests/test_overwrite.py
@@ -0,0 +1,83 @@
+import pathlib
+
+import pytest
+
+from zimit.zimit import run
+
+TEST_DATA_DIR = pathlib.Path(__file__).parent / "data"
+
+
+def test_overwrite_flag_behaviour(tmp_path):
+    zim_output = "overwrite-test.zim"
+    output_path = tmp_path / zim_output
+
+    # 1st run → creates file
+    result = run(
+        [
+            "--seeds",
+            "https://example.com",
+            "--warcs",
+            str(TEST_DATA_DIR / "example-response.warc"),
+            "--output",
+            str(tmp_path),
+            "--zim-file",
+            zim_output,
+            "--name",
+            "overwrite-test",
+        ]
+    )
+    assert result in (None, 100)
+    assert output_path.exists()
+
+    # 2nd run, no overwrite → should fail
+    with pytest.raises(SystemExit) as exc:
+        run(
+            [
+                "--seeds",
+                "https://example.com",
+                "--warcs",
+                str(TEST_DATA_DIR / "example-response.warc"),
+                "--output",
+                str(tmp_path),
+                "--zim-file",
+                zim_output,
+                "--name",
+                "overwrite-test",
+            ]
+        )
+    assert exc.value.code == 2
+
+    # 2nd run, no overwrite → should fail
+    with pytest.raises(SystemExit) as exc:
+        run(
+            [
+                "--seeds",
+                "https://example.com",
+                "--output",
+                str(tmp_path),
+                "--zim-file",
+                zim_output,
+                "--name",
+                "overwrite-test",
+            ]
+        )
+    assert exc.value.code == 2
+
+    # 3rd run, with overwrite → should succeed
+    result = run(
+        [
+            "--seeds",
+            "https://example.com",
+            "--warcs",
+            str(TEST_DATA_DIR / "example-response.warc"),
+            "--output",
+            str(tmp_path),
+            "--zim-file",
+            zim_output,
+            "--name",
+            "overwrite-test",
+            "--overwrite",
+        ]
+    )
+    assert result in (None, 100)
+    assert output_path.exists()