search.json

---
layout: none
---
{%- assign page_count = site.pages | size -%}
{%- assign index_count = 0 -%}
[
  {% for post in site.pages %}
  
    {% comment %} 
      (A) Here we can define files to exclude:
    {% endcomment %}
    {% unless post.url contains "404.html" 
                        or post.url contains "search.json" 
                        or post.url contains "search-script.js" 
                        or post.url contains "search-results.html" 
                        or post.url contains "search_original.json" 
                        or post.url contains "streamfunctions.html"
                        or post.url contains "html_template.html" %}

      {%- capture item_json -%}
      
        {% comment %} 
          (B) Here we define pages that have multiple components that we want to separate 
          in more sections. 
        {% endcomment %}
        {% if post.url contains "main_modules.html" or
              post.url contains "namelist.html" or 
              post.url contains "project.html" or
              post.url contains "setup_grid.html" or
              post.url contains "seeding_file.html" or
              post.url contains "read_field.html"%}
          
          {%- assign page_sections = post.content | split: '<div class="search-content"></div>' -%}
          {%- assign section_count = page_sections | size | minus: 1 -%}
          
          {%- assign section_keywords = post.keywords | split: 'break,' -%}

          {%- assign index_count = 0 -%}

          {% for single_section in page_sections offset: 1 %}

            {% if forloop.index != 1 or forloop.index != section_count %} 
              {%- assign index_count = index_count | plus: 1 -%}

              {%- assign section_title =  single_section | split: '</h2>' | slice: 0 | strip_html
                                                         | remove: '["' | remove: '"]' | replace: '\n', ' '| strip -%}

              {%- assign cleaned_section = single_section | strip_html 
                                                          | strip_newlines 
                                                          | replace: '    ', '' | replace: '    ', '' | replace: '  ', ' '  
                                                          | remove: section_title 
                                                          | url_encode | remove: '%22' | remove: '%23' | replace: '%3A', ' ' 
                                                          | replace: '%2C', ' ' | replace: '%28', ' ' | replace: '%29', ' ' | replace: '%5B', ' ' 
                                                          | replace: '%5D', ' ' | replace: '%7B', ' ' | replace: '%7D', ' ' | replace: '%3B', ' '
                                                          | replace: '%2F', ' ' | replace: '%5C', ' ' | replace: '%26', ' ' | replace: '%3C', ' ' 
                                                          | replace: '%3E', ' ' | replace: '%25', ' ' | replace: '%09', ' ' | replace: '%21', ' ' 
                                                          | replace: '%2A', ' ' | replace: '%2B', ' ' | replace: '%40', ' ' | replace: '%5E', ' ' 
                                                          | replace: '%7E', ' ' | replace: '%27', ' ' | replace: '%3F', ' ' | url_decode 
                                                          | remove: '  ' | remove: '  ' -%}
              

              {% if forloop.first %}
                {
                "title"    : "{{ post.title | escape }}",
                "category" : "{{ post.category }}",
                "keywords" : "{{ post.keywords | split: 'break' | slice: index_count }}",
                "url"      : "{{ site.baseurl }}{{ post.url }}",
                "content"  : "{{ cleaned_section | strip_newlines }}"

              } 
              {% else %}
                {
                  "title"    : "{{ post.title }}{{ ' - ' }}{{ section_title }}",
                  "category" : "{{ post.category }}",
                  "keywords" : "{{ post.keywords | split: 'break' | slice: index_count }}",
                  "url"      : "{{ site.baseurl }}{{ post.url }}{{ '#' }}{{  section_title | remove: '.' | downcase }}",
                  "content"  : "{{ cleaned_section | strip_newlines }}"

                } 
              {% endif %} ,


            
            {% endif %}
          
          {% endfor %}
          
        {% else %}
          {%- assign index_count = index_count | plus: 1 -%}

          {%- assign single_section = post.content | split: 'post-content' | slice: 1-%}

          {%- assign cleaned_section = single_section | strip_html | replace: '\n', ' '| strip_newlines | remove: '["\">' | remove: '<!-- ."]' 
                                                      | strip_newlines | remove: '    ' | replace: '  ', ' ' | replace: '  ', ' '
                                                       -%}
          {
            "title"    : "{{ post.title | escape }}",
            "category" : "{{ post.category }}",
            "keywords" : "{{ post.keywords | join: ', ' }}",
            "url"      : "{{ site.baseurl }}{{ post.url | default: 'index.html' }}",
            "content"  : "{{ cleaned_section }}"
          }
          {%- unless forloop.last -%},{%- endunless -%}
          
        {% endif %}
        
        
      {%- endcapture -%}
      
      {% comment %} 
        Output the JSON only if it's not empty, 
        and conditionally add a comma AFTER the item.
        NOTE: This still requires manually counting items or testing to fix the last comma.
      {% endcomment %}
      {{ item_json | strip }} 
      
      {% comment %} 
        This is the most reliable comma placement for complex loops.
        It checks if the current post is NOT the last one in the main list.
        It might still result in a trailing comma if the last item is filtered out. 
      {% endcomment %}
      

    {% endunless %}
  {% endfor %}
]