Call Trends API from Fortran with json-fortran

A Fortran job talks to Trends API with one POST to https://api.trendsapi.ai/api, a Bearer key, and a JSON body that names a mode. There is no official Fortran plugin. curl sends the request. json-fortran parses the envelope, then parses the body string a second time. Pulled on 2026-10-12, Google Search for fortran scored 8.0 for the week of 2026-10-10, down 38.46% from 13.0. fortran programming scored 36.0, up 16.13% from 31.0. gfortran scored 0.0, down 100.0% from 3.0. YouTube for fortran scored 71.0, up 77.5% from 40.0. Do not copy 36.0 onto Search 8.0. Do not treat gfortran Search 0.0 as the language token. The same helper that stores those rows also pulls a live Google Trends board; a rank-16 token that weekly Search printed at 5.0 versus 0.0 is documented in Art of Play Brooklyn live rank versus weekly Search.

How does a Fortran client call Trends API?

POST https://api.trendsapi.ai/api
Authorization: Bearer <api_key>
Content-Type: application/json

{"mode":"get_growth","source":"google search","keyword":"fortran","percent_growth":["7D","30D","3M","12M"]}

program trends_growth
  use json_module, only: json_file
  implicit none
  type(json_file) :: envelope, inner
  character(len=256) :: key
  character(len=:), allocatable :: body_str, cmd
  integer :: n
  logical :: found

  call get_environment_variable("TRENDSAPI_API_KEY", key, length=n)
  cmd = "curl -s -X POST https://api.trendsapi.ai/api" // &
        " --header 'Authorization: Bearer " // trim(key) // "'" // &
        " --header 'Content-Type: application/json'" // &
        " --data '{\"mode\":\"get_growth\",\"source\":\"google search\"," // &
        "\"keyword\":\"fortran\"," // &
        "\"percent_growth\":[\"7D\",\"30D\",\"3M\",\"12M\"]}'" // &
        " -o response.json"
  call execute_command_line(cmd)
  call envelope%load(filename="response.json")
  call envelope%get("body", body_str, found)
  ! body is a JSON string; parse it a second time
  call inner%load_from_string(body_str)
  call inner%print()
end program

The first mistake most clients make is treating the body field as an object after one parse. It is still a JSON string. Parse the envelope, then parse body a second time. Library docs:

https://github.com/jacobwilliams/json-fortran
https://curl.se/docs/manpage.html

Free tier allows 100 successful requests per month. Only HTTP 200 responses count against quota. Starter is 5,000, Pro is 25,000, Business is 100,000. There is no Fortran-specific source on this API. npm covers npmjs package names. python covers PyPI project names. A watchlist that needs Fortran package download counts has to pull those from somewhere else. A C sibling of this helper is on C Trends API. A C++ sibling is on C++ Trends API.

What did get_growth return for Fortran language tokens?

get_growth for fortran on google search tips on 2026-10-10. Store the exact strings. The language token, the programming phrase, the gfortran compiler token, and the json-fortran library token are different series.

get_time_series for fortran on Google Search closed at 8 on 2026-10-10, matching get_growth 8.0. Adjacent weeks: 13 on 2026-10-03, 45 on 2026-09-26, 43 on 2026-09-19, 64 on 2026-09-12. The series has 261 weekly points from 2021-10-16 through 2026-10-10. Peak in this pull is 100 on 2026-09-05, with 64 on 2026-09-12, 45 on 2026-09-26, and 26 on 2026-06-06. Tip 8 is not a new series high. 12M printed -42.86% versus 14.0. YTD printed 0.0% versus 8.0. 7D printed -38.46% versus 13.0. A job that alerts "Fortran is down" off a one-week -38.46% while YTD is flat at 8.0 versus 8.0 is reading a Search tip after a 100 spike, not the language.

fortran programming closed at 36, matching growth 36.0, up 16.13% from 31.0. Adjacent weeks: 31 on 2026-10-03, 83 on 2026-09-26, 63 on 2026-09-19, 84 on 2026-09-12. Peak in this pull is 100 on 2026-02-28, with 99 on 2026-09-05, 97 on 2026-05-30, and 94 on 2026-02-21. Tip 36 is not a new series high. YTD printed +9.09% versus 33.0 while 7D printed +16.13%. Do not copy 36.0 onto fortran 8.0. 7D signs disagree: programming phrase up, language token down. YTD signs also disagree: programming phrase +9.09% versus language token 0.0%. Magnitudes differ on 12M: programming phrase -38.98% versus language token -42.86%.

gfortran closed at 0, matching growth 0.0, down 100.0% from 3.0. Adjacent weeks: 3 on 2026-10-03, 30 on 2026-09-26, 34 on 2026-09-19, 55 on 2026-09-12. Peak in this pull is 100 on 2026-09-05, the same Saturday fortran Search also printed 100. After that shared spike the compiler series fell 100, 55, 34, 30, 3, 0. 12M was 0.0 versus 2.0 (-100.0%). YTD was 0.0 versus 1.0 (-100.0%). Close 0.0 sits next to fortran 8.0. Do not join those two closes. Do not treat the shared 2026-09-05 peak as proof the two series are one token.

json-fortran closed at 0.0 versus 0.0 on 7D, 12M, and YTD. data_quality was low. Report 0.0 versus 0.0. There is no Fortran package registry source. lapack closed at 6.0, down 33.33% from 9.0. 12M was 6.0 versus 16.0 (-62.5%). YTD was 6.0 versus 9.0 (-33.33%). A job that alerts "Fortran is down" off LAPACK 7D or json-fortran 0.0 is reading the math-library or JSON-library series.

Neighbors this session: c programming Search 43.0 versus 44.0 (-2.27% 7D), 12M 43.0 versus 62.0 (-30.65%), YTD 43.0 versus 44.0 (-2.27%). Do not copy 43.0 onto 8.0. c++ Search 75.0, down 9.64% from 83.0, 12M 75.0 versus 82.0 (-8.54%), YTD 75.0 versus 72.0 (+4.17%). rust programming Search 27.0, up 58.82% from 17.0, 12M 27.0 versus 32.0 (-15.62%), YTD 27.0 versus 30.0 (-10.0%). 7D signs diverge: rust programming +58.82% versus fortran -38.46%. go programming Search 16.0, down 30.43% from 23.0, 12M 16.0 versus 49.0 (-67.35%), YTD 16.0 versus 33.0 (-51.52%). matlab Search 100.0 versus 100.0 (0.0% 7D), 12M 100.0 versus 61.0 (+63.93%), YTD 100.0 versus 60.0 (+66.67%). Matlab at the Search ceiling is not fortran at 8.0. A MATLAB client that does the same double parse is on MATLAB Trends API. A Rust sibling is on Rust Trends API.

How do YouTube, News, and Wikipedia disagree with Search?

YouTube get_growth for fortran closed at 71.0 versus 40.0 (+77.5%). get_time_series closed at 71, matching growth. Adjacent weeks: 40 on 2026-10-03, 47 on 2026-09-26, 54 on 2026-09-19, 49 on 2026-09-12. Peak in this pull is 100 on 2025-11-01, with 98 on 2025-01-11, 96 on 2023-01-14, and 93 on 2022-02-26. Tip 71 is not a new YouTube series high. Search at 8 is not a new Search high. Search peaked at 100 on 2026-09-05; YouTube that same week printed 42. 12M YouTube was 71.0 versus 52.0 (+36.54%). YTD YouTube was 71.0 versus 38.0 (+86.84%). 3M YouTube was 71.0 versus 85.0 (-16.47%). 14D YouTube was 71.0 versus 47.0 (+51.06%). 30D YouTube was 71.0 versus 49.0 (+44.9%). 7D signs disagree: YouTube +77.5% versus Search -38.46%. Do not fill YouTube from Search 8.0.

Google News get_growth for fortran closed at 0.0 versus 0.0 on 7D, 14D, 30D, 3M, 12M, and YTD. data_quality was low. Keyword News 0.0 does not lead Search 8.0 and does not track YouTube 71.0. Absence of News coverage is a finding. Keyword News and the Google News Top News board are different feeds. Source docs for Search sit at Google Trends.

Wikipedia get_growth for fortran resolved to Fortran. 7D and 14D returned collapsed_range. September 2026 scored 13.7 with 20066 pageviews versus August 7.1 with 18752 pageviews (score +92.96%, volume +7.01%). 3M was 13.7 / 20066 versus June 6.1 / 18562, volume +8.1%. 12M was 13.7 / 20066 versus September 2025 40.5 / 25358, volume -20.87%. YTD volume was 20066 versus 26757 (-25.01%). Twenty thousand September views is a high-traffic language article. Fortran (programming language) on source wikipedia printed 14.5 score and 18 pageviews in September versus August 8.5 / 11 (score +70.59%, volume +63.64%). 3M volume was 18 versus 23 (-21.74%). 12M volume was 18 versus 13 (+38.46%). YTD volume was 18 versus 9 (+100.0%). Eighteen September views is not 20066. Compare Wikipedia on volume. Do not join 20066 pageviews onto Search 8.0. Do not join 18 pageviews onto Search 8.0. Source docs for Wikipedia sit at Wikipedia Trends.

How does a Fortran job pull a live board?

POST https://api.trendsapi.ai/api
Authorization: Bearer <api_key>
Content-Type: application/json

{"mode":"get_top_trends","type":"Google Trends","limit":25}

cmd = "curl -s -X POST https://api.trendsapi.ai/api" // &
      " --header 'Authorization: Bearer " // trim(key) // "'" // &
      " --header 'Content-Type: application/json'" // &
      " --data '{\"mode\":\"get_top_trends\",\"type\":\"Google Trends\"," // &
      "\"limit\":25}'" // &
      " -o board.json"
call execute_command_line(cmd)
call envelope%load(filename="board.json")
call envelope%get("body", body_str, found)
call inner%load_from_string(body_str)
call inner%print()

get_top_trends takes a feed type, not a keyword. Pair the live snapshot with sort=rank_change on 1d and 7d. On 2026-10-12T10:01:59Z the live Google Trends board ranked art of play brooklyn at 16 with 1d +34 from prev_rank 50. Weekly Search for that string closed at 5.0 versus 0.0, a first non-zero week. A helper that only stores 7D Search for fortran at 8.0 misses that board. Mode docs live at Docs.

What belongs in a Fortran watchlist?

Store fortran, fortran programming, gfortran, and json-fortran as separate Search series. Add YouTube fortran because the 0-100 close (71.0) is not Search 8.0 and 7D signs already disagree. Add Google News fortran because the series is 0.0 on every window in this pull. Add Wikipedia volume for the resolved Fortran title (20066 September views), not the 18-view Fortran (programming language) title and not the 0-100 score. Skip a Fortran-registry column: that source does not exist. Do not store lapack 6.0 as Fortran language adoption. Do not store gfortran 0.0 as the language token. Do not store matlab 100.0 as a Fortran close. Five Search keywords plus one YouTube series plus one live Google Trends board is 7 calls. Four weekly runs fit the free 100-request cap. Starter is 5,000, Pro 25,000, Business 100,000. Only successful 200 responses count.