Command line Czech news feed about the war in Ukraine (Deprecated)

ukreport.tcl

#!/usr/bin/env tclsh
# !! DEPRECATED !! THE NEWS ENDPOINT USED BY THIS SCRIPT HAS BEEN CLOSED !!

# Keep up with breaking news about Ukraine on the command line
# This Tcl script reports summaries of the most important events that
# took place since you last checked, scraped right from novinky.cz

# Dependencies: Tcl, curl
# Run with --raw, -r for a more parser-friendly format
# Run with --all, -a to list all the available news
# Run with --timef, -t [fmt] to specify time format (strftime)

# Jakub Sebek, May 2022, Public Domain

puts "As of August 2022, the news endpoint used by ukreport has been closed definitely."
if {[file isdirectory "~/.cache/ukreport"]} {
    file delete -force -- "~/.cache/ukreport"
    puts "Cache cleared."
}
exit 1

proc arg {a s {alt ""}} {
    set argi [lsearch -regexp $::argv "^(--$a$|-\[^-$s\]*$s)"]
    if {$alt != ""} {
        if {$argi == -1} { 
            return $alt
        } elseif {$argi != [llength $::argv]-1} { 
            return [lindex $::argv [expr {$argi + 1}]]
        } else { error "Argument $a is missing a value" }
    } else {
        return [expr {$argi != -1}]
    }
}

proc html2ansi {t} {
    regsub -all -- {<strong>} $t "\033\[1m" t
    regsub -all -- {</strong>} $t "\033\[0m" t
    regsub -all -- {<em>} $t "\033\[3m" t
    regsub -all -- {</em>} $t "\033\[0m"
}

proc paragraphs {ps mw} {
    set linew 0
    foreach p $ps {
        foreach w [split $p] {
            incr linew [string length $w]
            if {$linew >= $mw} {
                set linew [string length $w]
                puts ""
            }
            puts -nonewline "$w "
        }
        puts -nonewline "\n\n"
    }
}

proc plural {n str} {
    if {$n > 1} { return "$n ${str}s" } else { return "$n ${str}" }
}

set cachedir "~/.cache/ukreport"
file mkdir $cachedir
if {[file exists $cachedir/data]} {
    set cachefile [open $cachedir/data r]
    set cache [read -nonewline $cachefile]
    close $cachefile
} else {
    set cache [dict create "last_checked" [clock format 0 -format "%+"]]
}

set lastchecked [clock scan [dict get $cache "last_checked"] -format "%+"]
dict set cache "last_checked" [clock format [clock seconds] -format "%+"]

set url {https://www.novinky.cz/api/documents/40388819}
if {[catch {exec -- curl --silent --show-error --fail -- $url} html err]} { dict with err {
    puts ${-errorinfo}
    exit ${-code}
}}

set alltexts [list]
foreach {_ a} [regexp -inline -all -- {"texts"\s*:\s*\[([^]]*)\]} $html] {
    foreach {_ t _} [regexp -inline -all -- {"((\\"|[^"])*)"} $a] { 
        lappend alltexts [subst -nocommands -novariables $t] 
    }
}

set updates [list]
set dayshift 0
set lasttime [clock seconds]
foreach p $alltexts {
    if {[regexp -- {^\s*<strong>\s*(\d+)\D+(\d+)} $p _ h m]} {
        if {[info exists curr]} { lappend updates $curr }

        set t [clock scan "$h $m" -format "%H %M"]
        if {[clock add $t -$dayshift days] > $lasttime} { incr dayshift }

        set t [clock add $t -$dayshift days]
        regsub -- {^\s*<strong>\s*\d+\D+\d+[^\w<]*(</strong>[^\w<]*)?} $p {} p
        set curr [dict create "time" $t "content" [list $p]]
        set lasttime $t
    } else {
        dict lappend curr "content" $p
    }
}
if {[info exists curr]} { lappend updates $curr }

if {![arg raw r]} { puts [html2ansi "<em>It is [clock format [clock seconds] -format {%c}].</em>"] }
if {![arg all a]} {
    set lastupdatetime [dict get [lindex $updates 0] "time"]
    set updates [lmap u $updates {if {[dict get $u "time"] > $lastchecked} {expr {$u}} else break}]

    if {![arg raw r]} {
        if {[llength $updates] > 0} {
            puts [html2ansi "<em>[llength $updates] news since [clock format $lastchecked -format {%a %H:%M}].</em>"]
        } else {
            puts [html2ansi "<em>No news since [clock format $lastupdatetime -format {%a %H:%M}].</em>"]
        }
    }
}

foreach u [lreverse $updates] { dict with u {
    if {[arg raw r]} {
        puts [clock format $time -format [arg timef t {%a %H:%M}]]
        puts [join $content "\n"]
        puts ""
    } else {
        set minutediff [expr {([clock seconds] - $time) / 60}]
        set timestr [if {$minutediff < (60 * 3)} {
            if {$minutediff == 0} {
                expr {"Just now"}
            } elseif {$minutediff < 40} { 
                expr {"[plural $minutediff "minute"] ago"} 
            } else { 
                expr {"[plural [expr {($minutediff + 30) / 60}] "hour"] ago"}
            }
        } else {
            clock format $time -format [arg timef t {%a %H:%M}]
        }]
        puts [html2ansi "<em>\[$timestr\]</em>"]
        paragraphs [html2ansi $content] [arg width w 60]
    }
}}

puts [open $cachedir/data w] $cache