add importexport tests into contentcollector

more tests, WIP
2020-12-03 13:10:54 +01:00 · 2020-12-02 21:53:39 +01:00
45 changed files with 457 additions and 6 deletions
--- a/tests/backend/specs/api/importexport/brDontBreakOL.html
+++ b/tests/backend/specs/api/importexport/brDontBreakOL.html
@ -0,0 +1 @@
+<html><body><ol><li>should be 1</li><br/><li>should be 2</li><li>should be 3</li></ol><p></p></body></html>
--- a/tests/backend/specs/api/importexport/bulletListInOL.html
+++ b/tests/backend/specs/api/importexport/bulletListInOL.html
@ -0,0 +1 @@
+<html><body><ol><li>should be 1</li><ul><li>should be a bullet</li></ul><li>should be 2</li></ol><p></p></body></html>
--- a/tests/backend/specs/api/importexport/bulletsAndEmptyLines.html
+++ b/tests/backend/specs/api/importexport/bulletsAndEmptyLines.html
@ -0,0 +1 @@
+<html><body><ul class="list-bullet1"><li>bullet line 1</li><br/></ul><br/><ul class="list-bullet1"><li>bullet line 2</li><br/><li><ul class="list-bullet2"><li>bullet2 line 1</li></ul></li></ul><br/><br/><ul class="list-bullet1"><li><ul class="list-bullet2"><li>bullet2 line 2</li><li><br/></li></ul></li></ul></body></html>
--- a/tests/backend/specs/api/importexport/bulletsAndNonBulletLines.html
+++ b/tests/backend/specs/api/importexport/bulletsAndNonBulletLines.html
@ -0,0 +1 @@
+<html><body><ul class="list-bullet1"><li>bullet line 1</li>not bullet</ul>not bullet<ul class="list-bullet1"><li>bullet line 2</li><br/><ul class="list-bullet2"><li>bullet2 line 1</li></ul></ul></body></html>
--- a/tests/backend/specs/api/importexport/collapseSpacesAcrossNewlines.html
+++ b/tests/backend/specs/api/importexport/collapseSpacesAcrossNewlines.html
@ -0,0 +1,5 @@
+      <html><body>Need
+                    <span> more </span>
+                              space
+                                        <i>  s </i>
+                                                  !<br></body></html>
--- a/tests/backend/specs/api/importexport/collapseSpacesInsideElements.html
+++ b/tests/backend/specs/api/importexport/collapseSpacesInsideElements.html
@ -0,0 +1 @@
+<html><body>Need <span> more </span> space<i>  s </i> !<br></body></html>
--- a/tests/backend/specs/api/importexport/complexNest.html
+++ b/tests/backend/specs/api/importexport/complexNest.html
@ -0,0 +1 @@
+<!doctype html><html><body><ul class="bullet"><li>one</li><li>two</li><li>0</li><li>1</li><li>2<ul class="bullet"><li>3</li><li>4</li></ul></li></ul><ol class="number"><li>item<ol class="number"><li>item1</li><li>item2</li></ol></li></ol></body></html>
--- a/tests/backend/specs/api/importexport/dontDeleteSpaceInsideElements.html
+++ b/tests/backend/specs/api/importexport/dontDeleteSpaceInsideElements.html
@ -0,0 +1 @@
+<html><body>Need<span> more </span>space<i>s </i>!<br></body></html>
--- a/tests/backend/specs/api/importexport/ignoreAnyTagsOutsideBody.html
+++ b/tests/backend/specs/api/importexport/ignoreAnyTagsOutsideBody.html
@ -0,0 +1 @@
+<html><head><title>title</title><style></style></head><body>empty<br></body></html>
--- a/tests/backend/specs/api/importexport/indents.html
+++ b/tests/backend/specs/api/importexport/indents.html
@ -0,0 +1 @@
+<html><body><ul class="list-indent1"><li>indent1</li><li>indent1</li><li><ul class="list-indent2"><li>indent2</li><li>indent2</li></ul></li><li>indent1</li></ul><br></body></html>
--- a/tests/backend/specs/api/importexport/indentsAndEmptyLines.html
+++ b/tests/backend/specs/api/importexport/indentsAndEmptyLines.html
@ -0,0 +1 @@
+<html><body><ul class="list-indent1"><li>indented line 1</li><br/></ul><br/><ul class="list-indent1"><li>indented line 2</li><br/><li><ul class="list-indent2"><li>indent2 line 1</li></ul></li></ul><br/><br/><ul class="list-indent1"><li><ul class="list-indent2"><li>indent2 line 2</li><li><br/></li></ul></li></ul></body></html>
--- a/tests/backend/specs/api/importexport/indentsAndNewlines.html
+++ b/tests/backend/specs/api/importexport/indentsAndNewlines.html
@ -0,0 +1 @@
+<html><body><ul class="list-indent1"><li>indent1</li></ul><br/><ul class="list-indent1"><li>indent1</li><ul class="list-indent2"><li>indent2</li></ul></ul><br/><ul class="list-indent1"><ul class="list-indent2"><li>indent2</li></ul></ul></body></html>
--- a/tests/backend/specs/api/importexport/indentsAndNonIndentedLines.html
+++ b/tests/backend/specs/api/importexport/indentsAndNonIndentedLines.html
@ -0,0 +1 @@
+<html><body><ul class="list-indent1"><li>indent line 1</li>not indent</ul>not indent<ul class="list-indent1"><li>indent line 2</li><br/><li><ul class="list-indent2"><li>indent2 line 1</li></ul></li></ul></body></html>
--- a/tests/backend/specs/api/importexport/lineDoBreakInOl.html
+++ b/tests/backend/specs/api/importexport/lineDoBreakInOl.html
@ -0,0 +1 @@
+<html><body><ol><li>should be 1</li></ol><p>hello</p><ol><li>should be 1</li><li>should be 2</li></ol><p></p></body></html>
--- a/tests/backend/specs/api/importexport/lineDontBreakOL.html
+++ b/tests/backend/specs/api/importexport/lineDontBreakOL.html
@ -0,0 +1 @@
+<html><body><ol><li>should be 1</li><p></p><li>should be 2</li><li>should be 3</li></ol><p></p></body></html>
--- a/tests/backend/specs/api/importexport/lineWithSpacesPreceedsOL.html
+++ b/tests/backend/specs/api/importexport/lineWithSpacesPreceedsOL.html
@ -0,0 +1 @@
+<html><body><br><br><br> <ol><li>should be 1</li></ol></body></html>
--- a/tests/backend/specs/api/importexport/linebreakPreceedsOL.html
+++ b/tests/backend/specs/api/importexport/linebreakPreceedsOL.html
@ -0,0 +1 @@
+<html><body><br><ol><li>should be 1</li></ol></body></html>
--- a/tests/backend/specs/api/importexport/multiLineParagraph.html
+++ b/tests/backend/specs/api/importexport/multiLineParagraph.html
@ -0,0 +1,4 @@
+<html><body><p>
+а б в г ґ д е є ж з и і ї й к л м н о
+п р с т у ф х ц ч ш щ ю я ь</p>
+</body></html>
--- a/tests/backend/specs/api/importexport/multiLineParagraphWithPre.html
+++ b/tests/backend/specs/api/importexport/multiLineParagraphWithPre.html
@ -0,0 +1,8 @@
+<html><body><p> 
+а б в г ґ д е є ж з и і ї й к л м н о<pre>multiple
+lines
+in
+pre 
+</pre></p><p>п р с т у ф х ц ч ш щ ю я
+ь</p>
+</body></html>
--- a/tests/backend/specs/api/importexport/multipleIndentLevelAndStyles.html
+++ b/tests/backend/specs/api/importexport/multipleIndentLevelAndStyles.html
@ -0,0 +1 @@
+<html><body><ul class="list-indent1"><li>indent line 1</li></ul><br/><ul class="list-indent1"><li>indent line 2</li><li><ul class="list-indent2"><li>indent2 line 1</li></ul></li></ul><br/><ul class="list-indent1"><li><ul class="list-indent2"><li><ul class="list-indent3"><li><ul class="list-indent4"><li><span class="b s i u"><b><i><s><u>indent4 line 2 bisu</u></s></i></b></span></li><li><span class="b s"><b><s>indent4 line 2 bs</s></b></span></li><li><span class="u"><u>indent4 line 2 u</u></span><span class="u i s"><i><s><u>uis</u></s></i></span></li><li><ul class="list-indent5"><li><ul class="list-indent6"><li><ul class="list-indent7"><li><ul class="list-indent8"><li><span class="">foo</span></li><li><span class="b s"><b><s>foobar bs</b></s></span></li></ul></li></ul></li></ul></li></ul></li><li><ul class="list-indent5"><li>foobar</li></ul></li></ul></li></ul></li></ul></body></html>
--- a/tests/backend/specs/api/importexport/multipleNewLinesAtBeginning.html
+++ b/tests/backend/specs/api/importexport/multipleNewLinesAtBeginning.html
@ -0,0 +1 @@
+<html><body><br><br><p></p><p></p>first line<br><br>second line<br></body></html>
--- a/tests/backend/specs/api/importexport/multipleParagraphsWithSpacesPreceedsOL.html
+++ b/tests/backend/specs/api/importexport/multipleParagraphsWithSpacesPreceedsOL.html
@ -0,0 +1 @@
+<html><body><p> </p><p> </p><p> </p><ol><li>should be 1</li></ol></body></html>
--- a/tests/backend/specs/api/importexport/nestedOl.html
+++ b/tests/backend/specs/api/importexport/nestedOl.html
@ -0,0 +1 @@
+<html><body>a<ol><li>b<ol><li>c</li></ol></ol>notlist<p>foo</p></body></html>
--- a/tests/backend/specs/api/importexport/nestedOl2.html
+++ b/tests/backend/specs/api/importexport/nestedOl2.html
@ -0,0 +1 @@
+<html><body><ul><li>a<ol><li>b</li><li>c</li></ol></li></ul></body></html>
--- a/tests/backend/specs/api/importexport/nestedOrderedLi.html
+++ b/tests/backend/specs/api/importexport/nestedOrderedLi.html
@ -0,0 +1,14 @@
+<!doctype html>
+<html>
+  <body>
+    <ol>
+      <li>one</li>
+      <li>
+        <ol>
+          <li>1.1</li>
+        </ol>
+      </li>
+      <li>two</li>
+    </ol>
+  </body>
+</html>
--- a/tests/backend/specs/api/importexport/nonBreakingSpacesAfterNewlines.html
+++ b/tests/backend/specs/api/importexport/nonBreakingSpacesAfterNewlines.html
@ -0,0 +1 @@
+<html><body>something<br>&nbsp;&nbsp;&nbsp;something<br></body></html>
--- a/tests/backend/specs/api/importexport/nonBreakingSpacesAfterNewlinesP.html
+++ b/tests/backend/specs/api/importexport/nonBreakingSpacesAfterNewlinesP.html
@ -0,0 +1 @@
+<html><body>something<p></p>&nbsp;&nbsp;&nbsp;something<br></body></html>
--- a/tests/backend/specs/api/importexport/normalNewline.html
+++ b/tests/backend/specs/api/importexport/normalNewline.html
@ -0,0 +1 @@
+<html><body>line1<br>line2<br></body></html>
--- a/tests/backend/specs/api/importexport/ol.html
+++ b/tests/backend/specs/api/importexport/ol.html
@ -0,0 +1 @@
+<html><body><ol><li>a</li><li>b</li><li>c</li></ol><p>test</p></body></html>
--- a/tests/backend/specs/api/importexport/olWithNonDefaultStart.html
+++ b/tests/backend/specs/api/importexport/olWithNonDefaultStart.html
@ -0,0 +1 @@
+<html><body><ol start="6"><li>6</li><li>7</li></ol><ol start="4"><li>4</li><li>5</li></ol><ol class="list-number3" start="9"><li>0.0.9</li><li>0.0.10</li></ol></body></html>
--- a/tests/backend/specs/api/importexport/paragraphPreceedsOL.html
+++ b/tests/backend/specs/api/importexport/paragraphPreceedsOL.html
@ -0,0 +1 @@
+<html><body><p></p><ol><li>should be 1</li></ol></body></html>
--- a/tests/backend/specs/api/importexport/preIntroducesASpace.html
+++ b/tests/backend/specs/api/importexport/preIntroducesASpace.html
@ -0,0 +1,4 @@
+<html><body><p>
+    1
+<pre>preline
+</pre></p></body></html>
--- a/tests/backend/specs/api/importexport/spacesAfterNewline.html
+++ b/tests/backend/specs/api/importexport/spacesAfterNewline.html
@ -0,0 +1 @@
+<!doctype html><html><body>something<br>             something<br></body></html>
--- a/tests/backend/specs/api/importexport/spacesAfterNewlineP.html
+++ b/tests/backend/specs/api/importexport/spacesAfterNewlineP.html
@ -0,0 +1 @@
+<!doctype html><html><body>something<p></p>             something<br></body></html>
--- a/tests/backend/specs/api/importexport/spacesAtEndOfLine.html
+++ b/tests/backend/specs/api/importexport/spacesAtEndOfLine.html
@ -0,0 +1 @@
+<html><body>something            <br>             something<br></body></html>
--- a/tests/backend/specs/api/importexport/spacesAtEndOfLineP.html
+++ b/tests/backend/specs/api/importexport/spacesAtEndOfLineP.html
@ -0,0 +1 @@
+<html><body>something            <p></p>             something<br></body></html>
--- a/tests/backend/specs/api/importexport/stylingWithAttributesAndTags.html
+++ b/tests/backend/specs/api/importexport/stylingWithAttributesAndTags.html
@ -0,0 +1 @@
+<html><body>line<br/><span class="b s i u"><b><i><s><u>bold strikethrough italic underline</u></s></i></b></span><span> no style</span> no style<br></body></html>
--- a/tests/backend/specs/api/importexport/stylingWithAttributesWithoutTags.html
+++ b/tests/backend/specs/api/importexport/stylingWithAttributesWithoutTags.html
@ -0,0 +1 @@
+<html><body>line<br/><span class="b s i u">bold strikethrough italic underline</span><span> no style</span> no style<br></body></html>
--- a/tests/backend/specs/api/importexport/stylingWithFontAttributes.html
+++ b/tests/backend/specs/api/importexport/stylingWithFontAttributes.html
@ -0,0 +1 @@
+<html><body>line<br/><span style="font-style:italic"><span style="font-style:bold"><span style="font-style:underline"><span style="font-style:strikethrough">bold strikethrough italic underline</span></span></span></span><span> no style</span> no style</body></html>
--- a/tests/backend/specs/api/importexport/stylingWithSpanAndTagsWithoutAttributes.html
+++ b/tests/backend/specs/api/importexport/stylingWithSpanAndTagsWithoutAttributes.html
@ -0,0 +1 @@
+<html><body>line<br/><span><b><i><s><u>bold strikethrough italic underline</u></s></i></b></span><span> no style</span> no style<br></body></html>
--- a/tests/backend/specs/api/importexport/stylingWithTagsWithoutAttributes.html
+++ b/tests/backend/specs/api/importexport/stylingWithTagsWithoutAttributes.html
@ -0,0 +1 @@
+<html><body>line<br/><b><i><s><u>bold strikethrough italic underline</u></s></i></b><span> no style</span> no style<br></body></html>
--- a/tests/backend/specs/api/importexport/testP.html
+++ b/tests/backend/specs/api/importexport/testP.html
@ -0,0 +1 @@
+<html><body><p></p><p></p></body></html>
--- a/tests/backend/specs/api/importexport/ul.html
+++ b/tests/backend/specs/api/importexport/ul.html
@ -0,0 +1 @@
+<html><body><ul><li>a</li><li>b</li></ul><div>div</div><p>foo</p></body></html>
--- a/tests/backend/specs/api/importexport/ulIndented.html
+++ b/tests/backend/specs/api/importexport/ulIndented.html
@ -0,0 +1 @@
+<html><body><ul><li>a</li><ul><li>b</li></ul><li>a</li></ul><p>foo</p></body></html>
--- a/tests/backend/specs/contentcollector.js
+++ b/tests/backend/specs/contentcollector.js
@ -1,11 +1,19 @@
-const Changeset = require('../../../src/static/js/Changeset');
 const contentcollector = require('../../../src/static/js/contentcollector');
 const AttributePool = require('../../../src/static/js/AttributePool');
 const cheerio = require('../../../src/node_modules/cheerio');
-const util = require('util');

+/**
+ * The html of these tests is intentionally collapsed into a single line, except in special cases.
+ * All tests are duplicated in ./tests/backend/specs/api/importexport/ where they are properly indented.
+ *
+ * While importexport tests target the `setHTML` API endpoint, which is nearly identical to what happens
+ * when a user manually imports a document via the UI, the contentcollector tests don't use rehype to process
+ * the document. Rehype remove spaces and newĺines were applicable.
+ *
+ * Note the final <br> in every of the tests except when the body ends in a list.
+ */
 const tests = {
-  nestedLi: {
+  nestedOrderedLi: {
    description: 'Complex nested Li',
    html: '<!doctype html><html><body><ol><li>one</li><li><ol><li>1.1</li></ol></li><li>two</li></ol></body></html>',
    expectedLineAttribs: [
@ -89,7 +97,7 @@ const tests = {
    expectedText: ['a', '*b', '*c', 'notlist', 'foo'],
    noteToSelf: 'Ensure empty P does not induce line attribute marker, wont this break the editor?',
  },
-  nestedOl: {
+  nestedOl2: {
    description: 'First item being an UL then subsequent being OL will fail',
    html: '<html><body><ul><li>a<ol><li>b</li><li>c</li></ol></li></ul></body></html>',
    expectedLineAttribs: ['+1', '*0*1*2*3+1+1', '*0*4*2*5+1+1'],
@ -105,8 +113,375 @@ const tests = {
    noteToSelf: "<p></p>should create a line break but not break numbering -- This is what I can't get working!",
    disabled: true,
  },
+  brDontBreakOL: {
+    description: 'A single completely empty line break using <br/> within an ol should NOT reset count',
+    html: '<html><body><ol><li>should be 1</li><br/><li>should be 2</li><li>should be 3</li></ol><p></p></body></html>',
+    expectedLineAttribs: [],
+    expectedText: ['*should be 1', '*should be 2', '*should be 3'],
+    disabled: true,
+  },
+  normalNewline: {
+    description: 'A normal newline as <br> should work',
+    html: '<html><body>line1<br>line2<br></body></html>',
+    expectedLineAttribs: ['+5','+5'],
+    expectedText: ['line1', 'line2'],
+  },
+  indents: {
+    disabled:true,
+    description: 'Two indentations',
+    html: '<html><body><ul class="list-indent1"><li>indent1</li><li>indent1</li><li><ul class="list-indent2"><li>indent2</li><li>indent2</li></ul></li><li>indent1</li></ul><br></body></html>',
+    expectedLineAttribs: ['*0*1*2+1+7', '*0*1*2+1+7', '*0*3*2+1+7', '*0*3*2+1+7'],
+    expectedText: ['*indent1', '*indent1', '*indent2', '*indent2', '*indent1']
+  },
+  indentsAndNewlines: {
+    disabled: true, // does <ul><ul> which might not be right? instead: <ul><li><ul>
+    description: 'Indentations and newlines',
+    html: '<html><body><ul class="list-indent1"><li>indent1</li></ul><br/><ul class="list-indent1"><li>indent1</li><ul class="list-indent2"><li>indent2</li></ul></ul><br/><ul class="list-indent1"><ul class="list-indent2"><li>indent2</li></ul></ul></body></html>',
+    expectedLineAttribs: ['*0*1*2+1+7', '', '*0*1*2+1+7', '*0*3*2+1+7', '', '*0*1*2+1+7', '*0*3*2+1+7'],
+    expectedText: ['*indent1', '', '*indent1', '*indent2', '', '*indent1', '*indent2']
+  },
+  multipleIndentLevelAndStyles: {
+    disabled:true,
+    description: '8 levels of indents, newlines and attributes',
+    html: '<html><body><ul class="list-indent1"><li>indent line 1</li></ul><br/><ul class="list-indent1"><li>indent line 2</li><li><ul class="list-indent2"><li>indent2 line 1</li></ul></li></ul><br/><ul class="list-indent1"><li><ul class="list-indent2"><li><ul class="list-indent3"><li><ul class="list-indent4"><li><span class="b s i u"><b><i><s><u>indent4 line 2 bisu</u></s></i></b></span></li><li><span class="b s"><b><s>indent4 line 2 bs</s></b></span></li><li><span class="u"><u>indent4 line 2 u</u></span><span class="u i s"><i><s><u>uis</u></s></i></span></li><li><ul class="list-indent5"><li><ul class="list-indent6"><li><ul class="list-indent7"><li><ul class="list-indent8"><li><span class="">foo</span></li><li><span class="b s"><b><s>foobar bs</b></s></span></li></ul></li></ul></li></ul></li></ul></li><li><ul class="list-indent5"><li>foobar</li></ul></li></ul></li></ul></li></ul></body></html>',
+    expectedLineAttribs: [],
+    expectedText: ['*indent line 1', '*indent line 1']
+  },
+  bulletsAndEmptyLines: {
+    disabled: true,
+    description: 'Bullet lists that contain empty lines',
+    html: '<html><body><ul class="list-bullet1"><li>bullet line 1</li><br/></ul><br/><ul class="list-bullet1"><li>bullet line 2</li><br/><li><ul class="list-bullet2"><li>bullet2 line 1</li></ul></li></ul><br/><br/><ul class="list-bullet1"><li><ul class="list-bullet2"><li>bullet2 line 2</li><li><br/></li></ul></li></ul></body></html>',
+    expectedLineAttribs: [],
+    expectedText: ['*bullet line 1', '', '*bullet line 2', '', '*bullet2 line 1', '', '', '*bullet2 line 2', '']
+  },
+  indentsAndEmptyLines: {
+    disabled:true,
+    description: 'Indented lines that contain empty lines',
+    html: '<html><body><ul class="list-indent1"><li>indented line 1</li><br/></ul><br/><ul class="list-indent1"><li>indented line 2</li><br/><li><ul class="list-indent2"><li>indent2 line 1</li></ul></li></ul><br/><br/><ul class="list-indent1"><li><ul class="list-indent2"><li>indent2 line 2</li><li><br/></li></ul></li></ul></body></html>',
+    expectedLineAttribs: [],
+    expectedText: ['indented line 1', '', '', 'indented line 2', '', 'indent2 line 1']
+  },
+  bulletsAndNonBulletLines: {
+    disabled:true,
+    description: 'Bullet lists that contain not bullet lines',
+    html: '<html><body><ul class="list-bullet1"><li>bullet line 1</li>not bullet</ul>not bullet<ul class="list-bullet1"><li>bullet line 2</li><br/><ul class="list-bullet2"><li>bullet2 line 1</li></ul></ul></body></html>',
+    expectedLineAttribs: [],
+    expectedText: []
+  },
+  indentsAndNonIndentedLines: {
+    disabled: true,
+    description: 'Indented lines that contain non indented lines',
+    html: '<html><body><ul class="list-indent1"><li>indent line 1</li>not indent</ul>not indent<ul class="list-indent1"><li>indent line 2</li><br/><li><ul class="list-indent2"><li>indent2 line 1</li></ul></li></ul></body></html>',
+    expectedLineAttribs: [],
+    expectedText: ['*indent line 1', 'not indent', 'not indent', '*indent line 2', '', '*indent2 line 1']
+  },
+  olWithNonDefaultStart: {
+    disabled: true,
+    description: 'An ordered list that does not start with 1',
+    html: '<html><body><ol start="6"><li>6</li><li>7</li></ol><ol start="4"><li>4</li><li>5</li></ol><ol class="list-number3" start="9"><li>0.0.9</li><li>0.0.10</li></ol></body></html>',
+    expectedLineAttribs: [
+    ],
+    expectedText: ['*6', '*7', '*4', '*5', '*0.0.9', '*0.0.10'] //0.0. is wrong
+//totally wrong after import
+/**
+1. 6
+1. 7
+2. 4
+3. 5
+  3.0.1 0.0.9
+  3.0.1 0.0.10
+
+only backwards key (and probably return key) trigger renumbering, it is not triggered when entering in any of the lines:
+1. 6
+2. 7
+3. 4
+4. 5
+  4.0.1. 0.0.9
+  4.0.2. 0.0.10
+*/
+  },
+  stylingWithAttributesAndTags: {
+    description: 'Styling applied as separate tags and span attributes',
+    html: '<html><body>line<br/><span class="b s i u"><b><i><s><u>bold strikethrough italic underline</u></s></i></b></span><span> no style</span> no style<br></body></html>',
+    expectedLineAttribs: ['+4', '*0*1*2*3+z+i'],
+    expectedText: ['line', 'bold strikethrough italic underline no style no style']
+  },
+  stylingWithSpanAndTagsWithoutAttributes: {
+    description: 'Styling applied as separate tags inside a span',
+    html: '<html><body>line<br/><span><b><i><s><u>bold strikethrough italic underline</u></s></i></b></span><span> no style</span> no style<br></body></html>',
+    expectedLineAttribs: ['+4', '*0*1*2*3+z+i'],
+    expectedText: ['line', 'bold strikethrough italic underline no style no style']
+  },
+  stylingWithAttributesWithoutTags: {
+    disabled:true,
+    description: 'Styling applied as span attributes',
+    html: '<html><body>line<br/><span class="b s i u">bold strikethrough italic underline</span><span> no style</span> no style<br></body></html>',
+    expectedLineAttribs: ['+4', '*0*1*2*3+z+i'],
+    expectedText: ['line', 'bold strikethrough italic underline no style no style']
+  },
+  stylingWithTagsWithoutAttributes: {
+    description: 'Styling applied as separate tags',
+    html: '<html><body>line<br/><b><i><s><u>bold strikethrough italic underline</u></s></i></b><span> no style</span> no style<br></body></html>',
+    expectedLineAttribs: ['+4', '*0*1*2*3+z+i'],
+    expectedText: ['line', 'bold strikethrough italic underline no style no style']
+  },
+  stylingWithFontAttributes: {
+    disabled: true,
+    description: "Styling applied as font-style",
+    html: '<html><body>line<br/><span style="font-style:italic"><span style="font-style:bold"><span style="font-style:underline"><span style="font-style:strikethrough">bold strikethrough italic underline</span></span></span></span><span> no style</span> no style</body></html>',
+    expectedLineAttribs: [],
+    expectedText: ['abc']
+  },
+  dontDeleteSpaceInsideElements: { //BEGINNING AAND END
+    description: 'Preserve spaces on the beginning and end of a element',
+    html: '<html><body>Need<span> more </span>space<i>s </i>!<br></body></html>',
+    expectedLineAttribs: ['+f*0+2+1'],
+    expectedText: ['Need more spaces !']
+  },
+  collapseSpacesInsideElements: {
+    description: 'Preserve only on space when multiple are present',
+    html: '<html><body>Need <span> more </span> space<i>  s </i> !<br></body></html>',
+    expectedLineAttribs: ['+f*0'],
+    expectedText: ['Need more space s !'],
+    disabled: true
+  },
+  collapseSpacesAcrossNewlines: {
+    disabled:true,
+    description: 'Newlines and multiple spaces across newlines should be collapsed',
+    html: `
+      <html><body>Need
+          <span> more </span>
+          space
+          <i>  s </i>
+          !<br></body></html>`,
+    expectedLineAttribs: [],
+    expectedText: []
+  },
+  multipleNewLinesAtBeginning: {
+    disabled: true,
+    description: 'Multiple new lines at the beginning should be preserved',
+    html: '<html><body><br><br><p></p><p></p>first line<br><br>second line<br></body></html>',
+    expectedLineAttribs: [],
+    expectedText: []
+  },
+  spacesAfterNewline: {
+    description: 'Don\'t collapse spaces that follow a newline',
+    html: '<!doctype html><html><body>something<br>             something<br></body></html>',
+    expectedLineAttribs: ['+9', '+m'],
+    expectedText: ['something', '             something']
+  },
+  spacesAfterNewlineP: {
+    description: 'Don\'t collapse spaces that follow a newline',
+    html: '<!doctype html><html><body>something<p></p>             something<br></body></html>',
+    expectedLineAttribs: ['+9', '', '+m'],
+    expectedText: ['something', '', '             something']
+  },
+  spacesAtEndOfLine: {
+    description: 'Don\'t collapse spaces that follow a newline',
+    html: '<html><body>something            <br>             something<br></body></html>',
+    expectedLineAttribs: ['+l', '+m'],
+    expectedText: ['something            ', '             something']
+  },
+  spacesAtEndOfLineP: {
+    description: 'Don\'t collapse spaces that follow a newline',
+    html: '<html><body>something            <p></p>             something<br></body></html>',
+    expectedLineAttribs: ['+l', '', '+m'],
+    expectedText: ['something            ', '', '             something']
+  },
+  nonBreakingSpacesAfterNewlines: {
+    description: 'Don\'t collapse non-breaking spaces that follow a newline',
+    html: '<html><body>something<br>&nbsp;&nbsp;&nbsp;something<br></body></html>',
+    expectedLineAttribs: ['+9', '+c'],
+    expectedText: ['something', '   something']
+  },
+  nonBreakingSpacesAfterNewlinesP: {
+    description: 'Don\'t collapse non-breaking spaces that follow a newline',
+    html: '<html><body>something<p></p>&nbsp;&nbsp;&nbsp;something<br></body></html>',
+    expectedLineAttribs: ['+9', '', '+c'],
+    expectedText: ['something', '', '   something']
+  },
+  ignoreAnyTagsOutsideBody: {
+    description: 'Never add non-breaking spaces for elements outside body',
+    html: '<html><head><title>title</title><style></style></head><body>empty<br></body></html>',
+    expectedLineAttribs: ['+5'],
+    expectedText: ['empty']
+  },
+  linebreakPreceedsOL: {
+    description: 'A break before an ordered list should not introduce a list item',
+    html: '<html><body><br><ol><li>should be 1</li></ol></body></html>',
+    //expectedHTML: '<!DOCTYPE HTML><html><body><br><ol start="1" class="number"><li>should be 1</ol><br></body></html>',
+    expectedLineAttribs: ['', '*0*1*2*3+1+b'],
+    //expectedText: '\n\t1. should be 1\n\n'
+    expectedText: ['', '*should be 1']
+  },
+  paragraphPreceedsOL:{
+    description: 'An empty paragraph before an ordered list should not introduce a list item',
+    html: '<html><body><p></p><ol><li>should be 1</li></ol></body></html>',
+    //expectedHTML: '<!DOCTYPE HTML><html><body><br><ol start="1" class="number"><li>should be 1</ol><br></body></html>',
+    //expectedText: '\n\t1. should be 1\n\n'
+    expectedLineAttribs: ['', '*0*1*2*3+1+b'],
+    expectedText: ['', '*should be 1']
+  },
+  lineWithSpacesPreceedsOL:{
+    disabled:true,
+    description: 'A single line with a space right before an ordered list should not introduce a list item',
+    html: '<html><body><br><br><br> <ol><li>should be 1</li></ol></body></html>',
+    expectedLineAttribs: ['', '*0*1*2*3+1+b'],
+    expectedText: ['', '', ' ', '*should be 1']
+    //expectedHTML: '<!DOCTYPE HTML><html><body><br><br><br><ol start="1" class="number"><li>should be 1</ol><br></body></html>',
+    //expectedText: '\n\n\n\t1. should be 1\n\n'
+  },
+  multipleParagraphsWithSpacesPreceedsOL: {
+    description: 'Paragraphs with spaces right before an ordered list should not introduce a list item',
+    html: '<html><body><p> </p><p> </p><p> </p><ol><li>should be 1</li></ol></body></html>',
+    expectedLineAttribs: ['+1', '+1', '+1', '*0*1*2*3+1+b'],
+    expectedText: [' ', ' ', ' ', '*should be 1']
+    //expectedHTML: '<!DOCTYPE HTML><html><body><br><br><br><ol start="1" class="number"><li>should be 1</ol><br></body></html>',
+    //expectedText: '\n\n\n\t1. should be 1\n\n'
+  },
+  multiLineParagraph:{
+    description: "A paragraph with multiple lines should not loose spaces when lines are combined",
+    html: `<html><body><p>
+а б в г ґ д е є ж з и і ї й к л м н о
+п р с т у ф х ц ч ш щ ю я ь</p>
+</body></html>`,
+    expectedLineAttribs: [ '+1t' ],
+    expectedText: ["а б в г ґ д е є ж з и і ї й к л м н о п р с т у ф х ц ч ш щ ю я ь"]
+  },
+  multiLineParagraphWithPre:{
+    description: "lines in preformatted text should be kept intact",
+    html: `<html><body><p>
+а б в г ґ д е є ж з и і ї й к л м н о<pre>multiple
+lines
+in
+pre
+</pre></p><p>п р с т у ф х ц ч ш щ ю я
+ь</p>
+</body></html>`,
+    expectedLineAttribs: [ '+11', '+8', '+5', '+2', '+3', '+r' ],
+    expectedText: ['а б в г ґ д е є ж з и і ї й к л м н о', 'multiple', 'lines', 'in', 'pre', 'п р с т у ф х ц ч ш щ ю я ь']
+  },
+  preIntroducesASpace: {
+    description: "pre should be on a new line not preceeded by a space",
+    html: `<html><body><p>
+    1
+<pre>preline
+</pre></p></body></html>`,
+    expectedLineAttribs: [ '+6', '+7' ],
+    expectedText: ['    1 ', 'preline']
+  },
+  linesWithMoreThanOnceSpace: {
+    description: 'Multiple spaces should be preserved',
+    html: '<html><body>Text with  more   than    one space.<br></body></html>',
+    expectedLineAttribs: [ '+10' ],
+    expectedText: ['Text with  more   than    one space.']
+  },
+  'nonelistiteminlist #3620': {
+    //fails as a new list item is introduced
+    disabled:true,
+    description: 'Text content between <ul> and <li>',
+    html: '<html><body><ul>test<li>FOO</li></ul></body></html>',
+    expectedLineAttribs: ['*0*2+1+4', '*0*1*2+1+3'],
+    expectedText: ['test', '*FOO']
+  },
+  'whitespaceinUL #3620': {
+    //fails as a new list item is introduced
+    disabled:true,
+    description: 'Only spaces and no text between <ul> and <li>',
+    html: '<html><body><ul> <li>FOO</li></ul></body></html>',
+    expectedLineAttribs: ['*0*1*2+1+3'],
+    expectedText: ['*FOO']
+  },
+  whitespaceinBulletList: {
+    //fails as a new list item is introduced
+    disabled:true,
+    description: 'Only spaces and no text between <ul> and <li> in an bullet list',
+    html: '<!doctype html><html><body><ul class="bullet"> <li>FOO</li></ul></body></html>',
+    expectedLineAttribs: ['*0*1*2+1+3'],
+    expectedText: ['*FOO']
+  },
+  whitespaceinIntendedList: {
+    //fails as a new list item is introduced
+    disabled:true,
+    description: 'Only spaces and no text between <ul> and <li> in an indented list',
+    html: '<html><body><ul class="indent"> <li>FOO</li></ul></body></html>',
+    expectedLineAttribs: ['*0*1*2+1+3'],
+    expectedText: ['*FOO']
+  },
+  whitespaceinOrderedList: {
+    //fails as a new list item is introduced
+    disabled:true,
+    description: 'Only spaces and no text between <ol> and <li>',
+    html: '<html><body><ol> <li>FOO</li></ol></body></html>',
+    expectedLineAttribs: ['*0*1*2*3+1+3'],
+    expectedText: ['*FOO']
+  },
+  prefixcorrectlinenumber: {
+    description: 'A normal ordered list with two items',
+    html: '<html><body><ol><li>should be 1</li><li>should be 2</li></ol></body></html>',
+    expectedLineAttribs: ['*0*1*2*3+1+b', '*0*1*2*3+1+b'],
+    expectedText: ['*should be 1', '*should be 2']
+  },
+  prefixcorrectlinenumbernested: {
+    description: 'A normal nested ordered list but <ol><ol> without the second <ol> being inside <li>',
+    html: '<html><body><ol><li>should be 1</li><ol><li>foo</li></ol><li>should be 2</li></ol></body></html>',
+    expectedLineAttribs: ['*0*1*2*3+1+b', '*0*4*2*3+1+3', '*0*1*2*5+1+b'],
+    expectedText: ['*should be 1', '*foo', '*should be 2']
+  },
+  prefixcorrectlinenumbernested2: {
+    description: 'A normal nested ordered list with the second starting <ol> in it\'s own <li>',
+    html: '<html><body><ol><li>should be 1</li><li><ol><li>foo</li></ol></li><li>should be 2</li></ol></body></html>',
+    expectedLineAttribs: ['*0*1*2*3+1+b', '*0*4*2*3+1+3', '*0*1*2*5+1+b'],
+    expectedText: ['*should be 1', '*foo', '*should be 2']
+  },
+  prefixcorrectlinenumbernested3: {
+    description: 'A normal nested ordered list with <ol><li><ol>',
+    html: '<html><body><ol><li>should be 1<ol><li>foo</li></ol></li><li>should be 2</li></ol></body></html>',
+    expectedLineAttribs: ['*0*1*2*3+1+b', '*0*4*2*3+1+3', '*0*1*2*5+1+b'],
+    expectedText: ['*should be 1', '*foo', '*should be 2']
+  },
+//
+//  /*
+//  "prefixcorrectlinenumber when introduced none list item - currently not supported see #3450":{
+//    html: '<html><body><ol><li>should be 1</li>test<li>should be 2</li></ol></body></html>',
+//    expectedLineAttribs: '<!DOCTYPE HTML><html><body><ol start="1" class="number"><li>should be 1</li>test<li>should be 2</li></ol><br></body></html>',
+//    expectedText: '\t1. should be 1\n\ttest\n\t2. should be 2\n\n'
+//  }
+//  ,
+//  "newlinesshouldntresetlinenumber #2194":{
+//    html: '<html><body><ol><li>should be 1</li>test<li>should be 2</li></ol></body></html>',
+//    expectedLineAttribs: '<!DOCTYPE HTML><html><body><ol class="number"><li>should be 1</li>test<li>should be 2</li></ol><br></body></html>',
+//    expectedText: '\t1. should be 1\n\ttest\n\t2. should be 2\n\n'
+//  }
+//  */
+//  'nestedULsfollowedByNestedOLs': {
+//    html: `<!doctype html>
+//    <html><body>
+//    <ul class="bullet">
+//      <li>one</li>
+//      <li>two</li>
+//      <li>0</li>
+//      <li>1</li>
+//      <li>2
+//        <ul class="bullet">
+//          <li>3</li>
+//          <li>4</li>
+//        </ul>
+//      </li>
+//    </ul>
+//    <ol class="number">
+//      <li>item
+//        <ol class="number">
+//          <li>item1</li>
+//          <li>item2</li>
+//        </ol>
+//      </li>
+//    </ol></body></html>`,
+//    expectedLineAttribs: '<!DOCTYPE HTML><html><body><ul class="bullet"><li>one</li><li>two</li><li>0</li><li>1</li><li>2<ul class="bullet"><li>3</li><li>4</ul></li></ul><ol start="1" class="number"><li>item<ol start="2" class="number"><li>item1</li><li>item2</ol></li></ol><br></body></html>',
+//    expectedText: '\t* one\n\t* two\n\t* 0\n\t* 1\n\t* 2\n\t\t* 3\n\t\t* 4\n\t1. item\n\t\t1.1. item1\n\t\t1.2. item2\n\n'
+//  }
+}

-};

 describe(__filename, function () {
  for (const test in tests) {
@ -120,7 +495,7 @@ describe(__filename, function () {

      it(testObj.description, function (done) {
        const $ = cheerio.load(testObj.html); // Load HTML into Cheerio
-        const doc = $('html')[0]; // Creates a dom-like representation of HTML
+        const doc = $('body')[0]; // Creates a dom-like representation of HTML
        // Create an empty attribute pool
        const apool = new AttributePool();
        // Convert a dom tree into a list of lines and attribute liens
@ -130,6 +505,8 @@ describe(__filename, function () {
        const result = cc.finish();
        const recievedAttributes = result.lineAttribs;
        const expectedAttributes = testObj.expectedLineAttribs;
+//        console.warn(result)
+//        console.warn("the pool:",apool)
        const recievedText = new Array(result.lines);
        const expectedText = testObj.expectedText;
Author	SHA1	Message	Date
webzwo0i	6e33e43cbd	add importexport tests into contentcollector	2020-12-03 13:10:54 +01:00
webzwo0i	97df9b9afc	more tests, WIP	2020-12-02 21:53:39 +01:00
				`@ -0,0 +1 @@`
				`<html><body><ol><li>should be 1</li><br/><li>should be 2</li><li>should be 3</li></ol><p></p></body></html>`
				`@ -0,0 +1 @@`
				`<html><body><ol><li>should be 1</li><ul><li>should be a bullet</li></ul><li>should be 2</li></ol><p></p></body></html>`
				`@ -0,0 +1 @@`
				`<html><body><ul class="list-bullet1"><li>bullet line 1</li><br/></ul><br/><ul class="list-bullet1"><li>bullet line 2</li><br/><li><ul class="list-bullet2"><li>bullet2 line 1</li></ul></li></ul><br/><br/><ul class="list-bullet1"><li><ul class="list-bullet2"><li>bullet2 line 2</li><li><br/></li></ul></li></ul></body></html>`
				`@ -0,0 +1 @@`
				`<html><body>Need <span> more </span> space<i> s </i> !<br></body></html>`
				`@ -0,0 +1 @@`
				`<!doctype html><html><body><ul class="bullet"><li>one</li><li>two</li><li>0</li><li>1</li><li>2<ul class="bullet"><li>3</li><li>4</li></ul></li></ul><ol class="number"><li>item<ol class="number"><li>item1</li><li>item2</li></ol></li></ol></body></html>`
				`@ -0,0 +1 @@`
				`<html><body>Need<span> more </span>space<i>s </i>!<br></body></html>`
				`@ -0,0 +1 @@`
				`<html><head><title>title</title><style></style></head><body>empty<br></body></html>`
				`@ -0,0 +1 @@`
				`<html><body><ul class="list-indent1"><li>indent1</li><li>indent1</li><li><ul class="list-indent2"><li>indent2</li><li>indent2</li></ul></li><li>indent1</li></ul><br></body></html>`
				`@ -0,0 +1 @@`
				`<html><body><ol><li>should be 1</li></ol><p>hello</p><ol><li>should be 1</li><li>should be 2</li></ol><p></p></body></html>`