mech-dump treats all local files as HTML

simbabque · oalders · commit 47b9835fcf0a · 2022-07-25T11:34:37.000-04:00
Fixes #63 Closes #102
diff --git a/Changes b/Changes
@@ -20,6 +20,9 @@ Revision history for WWW::Mechanize
     [FIXED]
     - tick() can now handle checkboxes without a value (GH#331) (Jordan M Adler
       and Julien Fiegehenn)
+    [ENHANCEMENTS]
+    - mech_dump now treats all local files like HTML regardless of what it
+      thinks their content types are (GH#63) (Julien Fiegehenn)
 
     [ENHANCEMENTS]
     - set_fields() and submit_form(with_fields => ...) can now set multiple
diff --git a/script/mech-dump b/script/mech-dump
@@ -72,6 +72,10 @@ Options:
 The order of the options specified is relevant.  Repeated options
 get repeated dumps.
 
+C<mech-dump> will only work on HTML documents when used on remote URLs, but
+will assume any local file you pass it is HTML. If it is not, there won't be
+any usable results.
+
 Proxy settings are specified through the environment (e.g. C<http_proxy=http://proxy.my.place/>).
 See L<LWP::UserAgent> for details.
 
@@ -101,9 +105,11 @@ else {
 
 $mech->env_proxy();
 foreach my $uri (@uris) {
+  my $no_ct_check;
   if ( -e $uri ) {
       require URI::file;
       $uri = URI::file->new_abs( $uri )->as_string;
+      $no_ct_check = 1; # we don't have to check the content type
   }
 
   my $response = $mech->get( $uri );
@@ -115,7 +121,10 @@ foreach my $uri (@uris) {
       $response = $mech->get( $uri );
       $response->is_success or die "Can't fetch $uri with username and password\n", $response->status_line, "\n";
   }
-  $mech->is_html or die qq{$uri returns type "}, $mech->ct, qq{", not "text/html"\n};
+
+  unless ($no_ct_check) {
+      $mech->is_html or die qq{$uri returns type "}, $mech->ct, qq{", not "text/html"\n};
+  }
 
   foreach my $action (@actions ) {
       $action->( $mech );
diff --git a/t/html_file.txt b/t/html_file.txt
@@ -0,0 +1,7 @@
+<html>
+    <body>
+        <form name="text-form" action="http://localhost">
+            <input type="text" name="one" />
+        </form>
+    </body>
+</html>
diff --git a/t/mech-dump/mech-dump.t b/t/mech-dump/mech-dump.t
@@ -21,7 +21,7 @@ if ( $^O eq 'VMS' ) {
 
 # Simply use a file: uri instead of the filename to make this test
 # more independent of what URI::* thinks.
-my $source = 'file:t/google.html t/find_inputs.html';
+my $source = 'file:t/google.html t/find_inputs.html t/html_file.txt';
 
 my $perl;
 $perl = $1 if $^X =~ /^(.+)$/;
@@ -57,6 +57,9 @@ POST http://localhost/ [3rd_form]
   YourSister=                    (text)
   YourSister=                    (text)
   submit=Submit                  (submit)
+
+GET http://localhost [text-form]
+  one=                           (text)
 EOF
 } else {
     $expected = <<'EOF';
@@ -85,6 +88,9 @@ POST http://localhost/ [3rd_form]
   YourSister=                    (text)
   YourSister=                    (text)
   submit=Submit                  (submit)
+
+GET http://localhost [text-form]
+  one=                           (text)
 EOF
 }
 

-Original file line number
+Diff line change
@@ @@ -0,0 +1,7 @@ @@
 +<html>
 +    <body>
 +        <form name="text-form" action="http://localhost">
 +            <input type="text" name="one" />
 +        </form>
 +    </body>
 +</html>