1 #!/bin/bash

     2 

     3 XMLNS_H="http://www.w3.org/1999/xhtml" 

     4 

     5 # If we set xmlns_h="…", we can omit: --option xmlns_h "$XMLNS_H"

     6 # because XML namespaces can be provided either as an option or as an environmental variable.

     7 # Options have precedence.

     8 

     9 findFiles() {

    10 	find -print0;

    11 }

    12 

    13 fetchAttributes() {

    14 	relpipe-in-filesystem \

    15 		--parallel 8 \

    16 		--file name \

    17 		--streamlet xpath \

    18 			--option xmlns_h "$XMLNS_H" \

    19 			--option attribute '.' --option mode boolean --as 'valid_xml'  \

    20 			--option attribute 'namespace-uri()'         --as 'root_xmlns' \

    21 			--option attribute '/h:html/h:head/h:title'  --as 'title' \

    22 			--option attribute 'count(//h:h1)'           --as 'h1_count' \

    23 			--option attribute 'count(//h:h2)'           --as 'h2_count' \

    24 			--option attribute 'count(//h:h3)'           --as 'h3_count'

    25 }

    26 

    27 filterAndOrder() {

    28 	relpipe-tr-sql \

    29 		--relation "pages" \

    30 			"SELECT

    31 				name,

    32 				title,

    33 				h1_count,

    34 				h2_count,

    35 				h3_count

    36 			FROM filesystem WHERE root_xmlns = ?

    37 			ORDER BY h1_count + h2_count + h3_count DESC

    38 			LIMIT 5" \

    39 			--type-cast 'h1_count' integer \

    40 			--type-cast 'h2_count' integer \

    41 			--type-cast 'h3_count' integer \

    42 			--parameter "$XMLNS_H";

    43 }

    44 

    45 findFiles | fetchAttributes | filterAndOrder | relpipe-out-gui -title "Pages and titles"