Fixing parser to discard whitespace after formatting tags, to remove …

…extra required newlines. (#250)
openlawteam · Mar 26, 2020 · 0004d06 · 0004d06
1 parent c40fea0
commit 0004d06
Show file tree

Hide file tree

Showing 6 changed files with 57 additions and 27 deletions.
diff --git a/shared/src/main/scala/org/adridadou/openlaw/parser/template/BlockRules.scala b/shared/src/main/scala/org/adridadou/openlaw/parser/template/BlockRules.scala
@@ -167,29 +167,39 @@ trait BlockRules extends Parser with ExpressionRules with GlobalRules {
   }
 
   def centeredLine: Rule1[TemplateText] = rule {
-    capture(centered) ~> ((_: String) => TemplateText(List(Centered)))
+    capture(centered) ~ optional(ws) ~> (
+        (_: String) => TemplateText(List(Centered))
+    )
   }
 
   def rightLine: Rule1[TemplateText] = rule {
-    capture(right) ~> ((_: String) => TemplateText(List(RightAlign)))
+    capture(right) ~ optional(ws) ~> (
+        (_: String) => TemplateText(List(RightAlign))
+    )
   }
 
   def rightThreeQuartersLine: Rule1[TemplateText] = rule {
-    capture(rightThreeQuarters) ~> (
+    capture(rightThreeQuarters) ~ optional(ws) ~> (
         (_: String) => TemplateText(List(RightThreeQuarters))
     )
   }
 
   def pageBreak: Rule1[TemplateText] = rule {
-    capture(pagebreak) ~ "\n" ~> ((_: String) => TemplateText(List(PageBreak)))
+    capture(pagebreak) ~ optional(ws) ~> (
+        (_: String) => TemplateText(List(PageBreak))
+    )
   }
 
   def sectionBreak: Rule1[TemplateText] = rule {
-    capture(sectionbreak) ~> ((_: String) => TemplateText(List(SectionBreak)))
+    capture(sectionbreak) ~ optional(ws) ~> (
+        (_: String) => TemplateText(List(SectionBreak))
+    )
   }
 
   def indentLine: Rule1[TemplateText] = rule {
-    capture(indent) ~> ((_: String) => TemplateText(List(Indent)))
+    capture(indent) ~ optional(ws) ~> (
+        (_: String) => TemplateText(List(Indent))
+    )
   }
 
   def textPart: Rule1[TemplateText] = rule {

diff --git a/shared/src/main/scala/org/adridadou/openlaw/parser/template/MarkdownRules.scala b/shared/src/main/scala/org/adridadou/openlaw/parser/template/MarkdownRules.scala
@@ -9,27 +9,29 @@ trait MarkdownRules extends GlobalRules {
   }
 
   def centeredLine: Rule1[Seq[TextElement]] = rule {
-    capture(centered) ~> ((_: String) => Seq(Centered))
+    capture(centered) ~ optional(ws) ~> ((_: String) => Seq(Centered))
   }
 
   def rightLine: Rule1[Seq[TextElement]] = rule {
-    capture(right) ~> ((_: String) => Seq(RightAlign))
+    capture(right) ~ optional(ws) ~> ((_: String) => Seq(RightAlign))
   }
 
   def rightThreeQuartersLine: Rule1[Seq[TextElement]] = rule {
-    capture(rightThreeQuarters) ~> ((_: String) => Seq(RightThreeQuarters))
+    capture(rightThreeQuarters) ~ optional(ws) ~> (
+        (_: String) => Seq(RightThreeQuarters)
+    )
   }
 
   def pageBreak: Rule1[Seq[TextElement]] = rule {
-    capture(pagebreak) ~> ((_: String) => Seq(PageBreak))
+    capture(pagebreak) ~ optional(ws) ~> ((_: String) => Seq(PageBreak))
   }
 
   def sectionBreak: Rule1[Seq[TextElement]] = rule {
-    capture(sectionbreak) ~> ((_: String) => Seq(SectionBreak))
+    capture(sectionbreak) ~ optional(ws) ~> ((_: String) => Seq(SectionBreak))
   }
 
   def indentLine: Rule1[Seq[TextElement]] = rule {
-    capture(indent) ~> ((_: String) => Seq(Indent))
+    capture(indent) ~ optional(ws) ~> ((_: String) => Seq(Indent))
   }
 
   def twoStar: Rule0 = rule(strong ~ !twoStar)

diff --git a/...src/main/scala/org/adridadou/openlaw/parser/template/printers/XHtmlAgreementPrinter.scala b/...src/main/scala/org/adridadou/openlaw/parser/template/printers/XHtmlAgreementPrinter.scala
@@ -334,10 +334,6 @@ final case class XHtmlAgreementPrinter(
             })
 
           case FreeText(t: Text) =>
-            // This might be necessary for some cases, but has not been enabled yet
-            // Consume any following text elements which are only newlines
-            //val remaining = xs.dropWhile(_ === FreeText(Text("\n")))
-
             // Generate text output
             val innerFrag = text(t.str)
 

diff --git a/shared/src/test/scala/org/adridadou/openlaw/parser/OpenlawTemplateLanguageParserSpec.scala b/shared/src/test/scala/org/adridadou/openlaw/parser/OpenlawTemplateLanguageParserSpec.scala
@@ -393,7 +393,7 @@ class OpenlawTemplateLanguageParserSpec extends FlatSpec with Matchers {
       """.stripMargin
 
     val text2 =
-      """<div class="openlaw-paragraph paragraph-1"><p class="no-section">a small title</p></div><ul class="list-lvl-1"><li><div class="openlaw-paragraph paragraph-2"><p>1. I should have a ul tag.<br /></p></div></li><li><div class="openlaw-paragraph paragraph-3"><p>2. So should I.<br /></p></div></li><li><div class="openlaw-paragraph paragraph-4"><p>3. And I also.<br /></p></div></li></ul><div class="openlaw-paragraph paragraph-5"><p class="no-section"><hr class="section-break" /></p></div><div class="openlaw-paragraph paragraph-6"><p class="no-section"><br />But I should not!<br /><br />    </p></div>"""
+      """<div class="openlaw-paragraph paragraph-1"><p class="no-section">a small title</p></div><ul class="list-lvl-1"><li><div class="openlaw-paragraph paragraph-2"><p>1. I should have a ul tag.<br /></p></div></li><li><div class="openlaw-paragraph paragraph-3"><p>2. So should I.<br /></p></div></li><li><div class="openlaw-paragraph paragraph-4"><p>3. And I also.<br /></p></div></li></ul><div class="openlaw-paragraph paragraph-5"><p class="no-section"><hr class="section-break" /></p></div><div class="openlaw-paragraph paragraph-6"><p class="no-section">But I should not!<br /><br />    </p></div>"""
 
     val result = forPreview(text)
     resultShouldBe(result, text2)
@@ -408,7 +408,7 @@ class OpenlawTemplateLanguageParserSpec extends FlatSpec with Matchers {
         |more text""".stripMargin
 
     val text2 =
-      """<div class="openlaw-paragraph paragraph-1"><p class="no-section">some text</p></div><div class="openlaw-paragraph paragraph-2"><p class="no-section"><hr class="pagebreak" /></p></div><div class="openlaw-paragraph paragraph-3"><p class="no-section"><br />more text</p></div>"""
+      """<div class="openlaw-paragraph paragraph-1"><p class="no-section">some text</p></div><div class="openlaw-paragraph paragraph-2"><p class="no-section"><hr class="pagebreak" /></p></div><div class="openlaw-paragraph paragraph-3"><p class="no-section">more text</p></div>"""
 
     val result = forPreview(text)
     resultShouldBe(result, text2)
@@ -1346,14 +1346,37 @@ class OpenlawTemplateLanguageParserSpec extends FlatSpec with Matchers {
   it should "be able to break pages" in {
     val text =
       """first paragraph of text
-      |\pagebreak
-      |second paragraph of text""".stripMargin
+        |\pagebreak
+        |
+        |second paragraph of text""".stripMargin
+    resultShouldBe(
+      forReview(text),
+      """<p class="no-section">first paragraph of text<br /></p><p class="no-section"><hr class="pagebreak" /></p><p class="no-section">second paragraph of text</p>"""
+    )
+  }
+
+  it should "drop extraneous newlines in pagebreak tag" in {
+    val text =
+      """first paragraph of text
+        |\pagebreak
+        |second paragraph of text""".stripMargin
     resultShouldBe(
       forReview(text),
       """<p class="no-section">first paragraph of text<br /></p><p class="no-section"><hr class="pagebreak" /></p><p class="no-section">second paragraph of text</p>"""
     )
   }
 
+  it should "drop extraneous newlines in sectionbreak tag" in {
+    val text =
+      """first paragraph of text
+      |\sectionbreak
+      |second paragraph of text""".stripMargin
+    resultShouldBe(
+      forReview(text),
+      """<p class="no-section">first paragraph of text<br /></p><p class="no-section"><hr class="section-break" /></p><p class="no-section">second paragraph of text</p>"""
+    )
+  }
+
   it should "be able to indent lines" in {
     val text =
       """first paragraph of text

diff --git a/...test/scala/org/adridadou/openlaw/parser/template/printers/XHtmlAgreementPrinterSpec.scala b/...test/scala/org/adridadou/openlaw/parser/template/printers/XHtmlAgreementPrinterSpec.scala
@@ -95,7 +95,7 @@ class XHtmlAgreementPrinterSpec
     val html = XHtmlAgreementPrinter(false)
       .printParagraphs(agreement.right.value.paragraphs)
       .print
-    html shouldBe """<p class="no-section align-right"> <strong>[[Company Name]]</strong></p>"""
+    html shouldBe """<p class="no-section align-right"><strong>[[Company Name]]</strong></p>"""
   }
 
   it should "handle right aligned underlined section" in {
@@ -104,7 +104,7 @@ class XHtmlAgreementPrinterSpec
     val html = XHtmlAgreementPrinter(false)
       .printParagraphs(agreement.right.value.paragraphs)
       .print
-    html shouldBe """<p class="no-section align-right"> <u>[[Company Name]]</u></p>"""
+    html shouldBe """<p class="no-section align-right"><u>[[Company Name]]</u></p>"""
   }
 
   it should "handle right 3/4 aligned section" in {
@@ -113,7 +113,7 @@ class XHtmlAgreementPrinterSpec
     val html = XHtmlAgreementPrinter(false)
       .printParagraphs(agreement.right.value.paragraphs)
       .print
-    html shouldBe """<p class="no-section align-right-three-quarters"> <strong>[[Company Name]]</strong></p>"""
+    html shouldBe """<p class="no-section align-right-three-quarters"><strong>[[Company Name]]</strong></p>"""
   }
 
   it should "handle agreements with multiple centered sections" in {
@@ -133,7 +133,7 @@ class XHtmlAgreementPrinterSpec
     val html = XHtmlAgreementPrinter(false)
       .printParagraphs(agreement.right.value.paragraphs)
       .print
-    html shouldBe """<p class="no-section align-center"> <strong>BYLAWS</strong><br /> <strong>OF</strong><br /> <strong>[[Company Name]]</strong><br /> (A DELAWARE CORPORATION)<br /></p><ul class="list-lvl-1"><li><p>1.  <strong>Offices</strong></p><ul class="list-lvl-2"><li><p>(a)  <strong>Registered Office</strong>.  The registered office of the corporation in the State of Delaware shall be [[Registered Agent Address]], and the name of the registered agent of the corporation in the State of Delaware at such address is [[Registered Agent Name]].<br /></p></li></ul></li></ul>"""
+    html shouldBe """<p class="no-section align-center"><strong>BYLAWS</strong><br /><strong>OF</strong><br /><strong>[[Company Name]]</strong><br />(A DELAWARE CORPORATION)<br /></p><ul class="list-lvl-1"><li><p>1.  <strong>Offices</strong></p><ul class="list-lvl-2"><li><p>(a)  <strong>Registered Office</strong>.  The registered office of the corporation in the State of Delaware shall be [[Registered Agent Address]], and the name of the registered agent of the corporation in the State of Delaware at such address is [[Registered Agent Name]].<br /></p></li></ul></li></ul>"""
 
   }
 

diff --git a/shared/src/test/scala/org/adridadou/openlaw/vm/OpenlawExecutionEngineSpec.scala b/shared/src/test/scala/org/adridadou/openlaw/vm/OpenlawExecutionEngineSpec.scala
@@ -2298,8 +2298,7 @@ class OpenlawExecutionEngineSpec extends FlatSpec with Matchers {
 
   it should "handle missing values in a structure" in {
     val template =
-      compile("""
-                 |\centered **Test Agreement - structure**
+      compile("""\centered **Test Agreement - structure**
  |
  |# Structure definition
  |[[Contestant Emergency Contact: Structure(
@@ -2330,7 +2329,7 @@ class OpenlawExecutionEngineSpec extends FlatSpec with Matchers {
       case Success(result) =>
         result.state shouldBe ExecutionFinished
         val text = parser.forReview(result.agreements.head)
-        text shouldBe "<p class=\"no-section\"><br /> <strong>Test Agreement - structure</strong></p><p class=\"no-section\"># Structure definition<br /></p><p class=\"no-section\"># Structure type var<br /></p><p class=\"no-section\"><strong>Emergency Contact</strong><br />Name: test<br />Age: [[Medical Contact]]<br />DOB: [[Medical Contact]]<br />Address: [[Medical Contact]]</p><p class=\"no-section\"><br />              </p>"
+        text shouldBe "<p class=\"no-section align-center\"><strong>Test Agreement - structure</strong></p><p class=\"no-section\"># Structure definition<br /></p><p class=\"no-section\"># Structure type var<br /></p><p class=\"no-section\"><strong>Emergency Contact</strong><br />Name: test<br />Age: [[Medical Contact]]<br />DOB: [[Medical Contact]]<br />Address: [[Medical Contact]]</p><p class=\"no-section\"><br />              </p>"
       case Failure(_, message) =>
         fail(message)
     }