-I/--input-dir defaulting to "downloaded" in extract/scrape

Yomguithereal · Yomguithereal · commit 39c98a89d140 · 2024-09-05T15:12:40.000+02:00
Fix #974
diff --git a/docs/cli.md b/docs/cli.md
@@ -868,6 +868,7 @@ Optional Arguments:
   -g, --glob                    Will interpret given paths as glob patterns to
                                 resolve if given.
   -I, --input-dir INPUT_DIR     Directory where the HTML files are stored.
+                                Defaults to `downloaded`.
   --mimetype-column MIMETYPE_COLUMN
                                 Name of the CSV column containing file mimetype.
                                 Defaults to `mimetype`.
@@ -1197,6 +1198,7 @@ Optional Arguments:
   -g, --glob                    Will interpret given paths as glob patterns to
                                 resolve if given.
   -I, --input-dir INPUT_DIR     Directory where the HTML files are stored.
+                                Defaults to `downloaded`.
   --mimetype-column MIMETYPE_COLUMN
                                 Name of the CSV column containing file mimetype.
                                 Defaults to `mimetype`.
diff --git a/minet/cli/extract/__init__.py b/minet/cli/extract/__init__.py
@@ -1,6 +1,7 @@
 from casanova import IndexedResumer
 
 from minet.cli.argparse import command
+from minet.cli.constants import DEFAULT_CONTENT_FOLDER
 
 
 def resolve_arguments(cli_args):
@@ -101,6 +102,7 @@ def resolve_arguments(cli_args):
         {
             "flags": ["-I", "--input-dir"],
             "help": "Directory where the HTML files are stored.",
+            "default": DEFAULT_CONTENT_FOLDER
         },
         {
             "flags": ["-p", "--processes"],
diff --git a/minet/cli/scrape/__init__.py b/minet/cli/scrape/__init__.py
@@ -1,4 +1,5 @@
 from minet.cli.argparse import command
+from minet.cli.constants import DEFAULT_CONTENT_FOLDER
 
 
 def resolve_arguments(cli_args):
@@ -126,6 +127,7 @@ def resolve_arguments(cli_args):
         {
             "flags": ["-I", "--input-dir"],
             "help": "Directory where the HTML files are stored.",
+            "default": DEFAULT_CONTENT_FOLDER
         },
         {
             "flags": ["-p", "--processes"],

Original file line number	Diff line number	Diff line change
`@@ -1,6 +1,7 @@`
`1`	`1`	`from casanova import IndexedResumer`
`2`	`2`
`3`	`3`	`from minet.cli.argparse import command`
	`4`	`+from minet.cli.constants import DEFAULT_CONTENT_FOLDER`
`4`	`5`
`5`	`6`
`6`	`7`	`def resolve_arguments(cli_args):`
`@@ -101,6 +102,7 @@ def resolve_arguments(cli_args):`
`101`	`102`	`{`
`102`	`103`	`"flags": ["-I", "--input-dir"],`
`103`	`104`	`"help": "Directory where the HTML files are stored.",`
	`105`	`+ "default": DEFAULT_CONTENT_FOLDER`
`104`	`106`	`},`
`105`	`107`	`{`
`106`	`108`	`"flags": ["-p", "--processes"],`
Original file line number	Diff line number	Diff line change
`@@ -1,4 +1,5 @@`
`1`	`1`	`from minet.cli.argparse import command`
	`2`	`+from minet.cli.constants import DEFAULT_CONTENT_FOLDER`
`2`	`3`
`3`	`4`
`4`	`5`	`def resolve_arguments(cli_args):`
`@@ -126,6 +127,7 @@ def resolve_arguments(cli_args):`
`126`	`127`	`{`
`127`	`128`	`"flags": ["-I", "--input-dir"],`
`128`	`129`	`"help": "Directory where the HTML files are stored.",`
	`130`	`+ "default": DEFAULT_CONTENT_FOLDER`
`129`	`131`	`},`
`130`	`132`	`{`
`131`	`133`	`"flags": ["-p", "--processes"],`