[general]
name=Tembuku - Feeder Tap Designer
qgisMinimumVersion=3.22
qgisMaximumVersion=4.99
version=1.4.4
author=Joseph Edward Timothy Siahaan
email=edwardtimoo@gmail.com
description=FTTH/GPON feeder design in QGIS: for each new FDT, reuse an existing feeder with a free tube or route a new one from the POP - tap point, cable route, optical budget and the reason - delivered as KMZ in the feeder compilation layout.
about=Tembuku turns the first stage of FTTH/GPON feeder design into a one-folder task in QGIS. Point it at a folder of map data (cable inventory, feeder design compilation, the FDT points to design, POP list) as KMZ/KML, ZIP, GeoJSON, CSV/Excel or Shapefile/GeoPackage: Tembuku recognises each file from what is inside it, not from its name, says why, lists and corrects small typing mistakes in folder names, FDT codes and coordinates, and stops with a clear message when a required file is missing or unsafe. For every FDT it (1) checks the free tubes of nearby feeders and whether each feeder really runs toward the FDT, (2) picks the tap point and draws the new cable along existing poles and roads (OpenStreetMap, cached), (3) checks the GPON optical budget with the maximum values of current ITU-T standards and (4) decides between reusing a feeder and building a new one from the POP, with the reason written out. Results: KMZ files in the compilation folder layout with automatic F/T numbering, styled QGIS layers, an interactive offline HTML report, a CSV summary and an accuracy test against designs already approved (17 of 17 decisions matched on the Bitung and Tomohon designs used to calibrate the rules). Click an FDT on the map for an answer in about a second, or choose another candidate in the results panel. Basemaps show only current imagery with a known date: OpenStreetMap, detailed satellite imagery with the capture date at the map centre and an age warning, and the newest cloud-free Copernicus Sentinel-2 mosaic for the whole of Indonesia. Every value can be changed and names its source; no company names are built in (optional company labels stay in a local file). Large KMZ files are indexed once per work area and read safely (no extraction, size limits, no DTD). No extra Python packages. Interface in English; reports in Indonesian. Licence GPL v3 or later with section 7 terms (author attribution kept, modified versions renamed and marked, no right to the Tembuku name or logo); official releases are digitally signed by the author (release key fingerprint 92C4-D078-D1FC-1C5B, the same as Seranta and Seuntai; any other key is not an official Tembuku release).
category=Vector
icon=icon.png
tags=ftth,basemap,satellite,gpon,fiber,fibre optic,fttx,feeder,telecommunication,network design,routing,kmz,colocation,osm
homepage=https://github.com/jetstimo/tembuku
tracker=https://github.com/jetstimo/tembuku/issues
repository=https://github.com/jetstimo/tembuku
hasProcessingProvider=yes
experimental=False
deprecated=False
server=False
changelog=1.4.4 Engineering rules of 11 Nov 2026 checked line by line, against credible technical sources and against 213 approved designs in four cities; two rules are kept configurable but no longer applied literally by default, because the data and the sources show they cost money without improving the design. (1) Tube group by distance (6 km): with the rule applied literally, only 36 percent of F/T folders matched the designers and 563 km of new cable was proposed; with it off, 54 percent and 425 km (designers: 495 km). Optical loss depends on route length, not on fibre count (ITU-T G.652, G.984.2), so the reservation has no optical basis. A 48-fibre branch over 6 km is still rejected and a branch whose fibre path is over 6 km is a 96-fibre cable. (2) Taps at existing slack: preferred when the new cable is at most 600 m (or 15 percent) longer, instead of always. A closure can be built mid-span (manufacturers' mid-span access practice) and is marked 'please check'; reaching a slack with kilometres of extra cable is dearer than one closure. Settings measured: 300 m 421 km / 99 checks, 600 m 425 km / 92 checks and the highest F/T match, 1000 m 434 km / 81 checks, always 481 km / 60 checks. KMZ follows the APD template more closely: the exact slack or closure coordinate starts the new cable, Colo Pole lines take the colour of their cable size, closure 96-144 uses the Closure 96 icon, hanger and OLT icons follow the template, 9 m crossing poles are written in the Tiang 9 Meter folder, the New Infra folder is no longer added, every new cable carries a plain-language summary (decision, F/T, tube, new cable, share on existing poles, optical loss, poles), and the POP folder no longer ends in '.kmz'. A team icon folder (for example 'ICON APD FTTH') next to the data folder is found automatically. The template check now works at any folder depth. Fix: with a feeder tube record table (Excel/CSV) in the data folder, the tube of an FDT that is designed again (and of every design hidden in the accuracy test) stayed marked as used, so the next tube or even a new feeder was proposed; on the Palembang POP 3 data with its tube record the same F/T folder went from 2 to 31 of 60 and the same decision from 41 to 48 of 60. Faster: nearest-point and corridor checks no longer repeat work, with the same results. 1.4.3 Better and cheaper decisions, checked on 213 approved designs in four cities (Bengkulu, Gresik, Palembang, Serang), and results written in plain words. Tube choice: Tembuku now takes the lowest free tube of the feeder, as the designers do, instead of reserving T9-T12 for FDTs within 6 km; the 6 km limit for 48-fibre cables still applies and the old behaviour can be switched on under Advanced. Tap points: a tap at an existing slack, closure or cable end (end-to-end) is preferred when its new cable is at most 300 m (or 15 percent) longer, because a cable cannot be opened mid-span without spare cable; an FDT right next to a feeder is still joined near the FDT. A tap that is not at such a point, and every proposed tube number not yet recorded in the compilation, is marked 'please check', as the engineering rules ask. Among equal candidates the one with more of its new cable on existing poles is preferred. A closure proposed for one FDT can be shared by the next FDTs of the same run. Routes avoid footways, pedestrian streets and motorways (configurable). 9 m crossing poles are also placed where the route crosses residential and unclassified roads, and hangers are moved to the nearest proposed pole. Every result now starts with one plain sentence: what to build, which F/T folder, how much new cable, and what a new feeder would need instead; reasons for rejected candidates are written in everyday words. Results on the 213 designs, against 1.4.2: same decision 82 percent (unchanged), same F/T folder 37 to 53 percent, route match median 42 to 73 percent, and total proposed new cable 523 to 420 km (20 percent less). 1.4.2Four data-reading fixes found by running four city compilations (213 designs) on 12 Nov 2026; the design rules and the route limit of 1.24 are unchanged. (1) Folder names that hold several tubes are now read ('F8_T1,T2,T3', 'F3_T3_T4_T5', 'F1_T10,11') as are folders without a cluster ID ('F1_T1_APD Feeder_...'); before, 62 of 342 folders were skipped and their tubes counted as free. (2) An FDT that is already in the compilation is designed again with its earlier design hidden, so it no longer taps its own cable (0 m of new cable) and its old tube is not counted twice. (3) The tube book CSV lists every POP and feeder of the compilation, also when the POP folder and the POP point have different names. (4) A POP folder without a POP/OLT point gets its location from the start of its 144-fibre feeder cables and is reported under 'please check', instead of being measured from a POP 7-14 km away. Accuracy test on the four compilations: same decision 76 to 82 percent, same F/T folder 25 to 37 percent. 1.4.1 More of the 11 Nov 2026 engineering rules. New feeder: the 144-fibre cable now ends in the slack window of its feeder number (F01 100-250 m, F02 150-300 m, 50 m further per feeder, never over 750 m). An existing closure is used as a tap point only while it has a free cable hole (one input, at most five outputs). New-infra stretches get proposed poles in the KMZ: 7 m poles every 50 m and 9 m poles on both sides of a main-road crossing (can be switched off under Advanced). Folders named Drop in a compilation are ignored. A new file TEMBUKU_BUKU_TUBE lists, per POP and feeder, the tubes in use, the free tubes of the 96F and 48F groups and the proposals of this run. FDTs designed in one run now see each other's proposed tube and cable, so two FDTs never get the same tube. 1.4.0 Smarter and steadier decisions in every region, following the engineering rules of 11 Nov 2026. Free tubes now come from the feeder design compilation alone, read from its folder hierarchy: one feeder is one 144-fibre trunk of 12 tubes, the T number of a folder is the tube it uses, T1-T8 go through the 96-fibre cable and T9-T12 through 48-fibre branches from the closure 144 at the POP; Tembuku proposes the lowest free tube of the right group (6 km rule) and writes it as the T number and in the FDT fields F (SP1) / T (SP1). Cable-inventory cables are existing pole routes only (they can be allowed again under Advanced). A new 48-fibre branch from a feeder's closure 144 is now a candidate, accepted only when its route follows that feeder's own corridor. Tap points move to an existing slack or closure of the feeder when one is within 10 m or close by; an existing closure is reused, otherwise a new closure is placed exactly at the slack. Routes follow roads: straight helper lines are used only where no road or pole route exists. The KMZ adds hanger/slack points every 300 m along the new cable and the OTB Existing point. Fixes: the POP folder is no longer mistaken for the file name (which gave F1_T1 in some regions); feeders sharing a pole line no longer count each other's FDTs; one physical cable is one candidate; the optical budget counts real splice points; earlier Tembuku result KMZ files in the data folder are skipped. Equal candidates are ordered by a fixed rule (existing slack first, then the lowest feeder number) and listed for review. On the Bengkulu POP 4 case the result equals the designer's (F2_T9, 80 percent of the route identical); the leave-one-out test on the same compilation went from 12 to 15 of 19 equal decisions, 4 to 10 equal F/T folders, and runs three times faster. 1.3.2 Speed: reading a large cable inventory for a new work area is about three times faster (column names are recognised once instead of for every cable), and repeated runs on the same data start in a few seconds (folder-name checks are remembered). Result files: the design KMZ is named like the feeder design files in the compilation (cluster ID, APD Feeder KMZ, area, date; the FDT code when those are missing) and the summary lists the file of each FDT. Each FDT now gets two KMZ files - the design KMZ holds only the template folders (nothing extra for a bill-of-materials tool to misread), and a companion file ending in _ANALISIS holds the review material (candidates, tap points, helper lines, stretches that need new poles). Lines in the Colo Pole folders take the colour of their cable size with the template line width, and on a new feeder they are split between the first cable and the rest instead of being repeated. Your own icons are recognised by what their file names start with (FDT, Closure 48, Closure 96, Closure 288, OTB, OLT, Spare) and stored in the KMZ under neutral names. New notes, decisions unchanged: a result whose signal loss is within 1 dB of the limit is flagged to be measured on site; the accuracy test names designs that cannot be tested fairly because their feeder has since been built and is now in the inventory; a feeder tube record is listed in the file table. 1.3.1 Older compilation layouts are now read in full: cables kept in "Net_Cables" folders (named like "48F 1234") are read as new cables with their fibre count, and "Colo Cable" lines kept in "Ket.Infra" folders are read as the existing feeder used - before, most of these lines were treated as notes, so feeders already designed were missed. A cable from the inventory that is in fact a feeder already in the compilation (same cluster ID in its name, or the tap point lies on that feeder) now gets that feeder number and the next T number instead of a new feeder number. Area names from folder names written "APD Feeder_<area>" no longer repeat the words "APD Feeder"; the template check no longer warns when two route folders have the same name. Faster first run in a new area: a nationwide POP list no longer widens the work area (only POPs near your FDTs and compilation count), and an inventory file that has been read once is skipped at once for areas where it holds no cable. KMZ files of two FDTs with the same code no longer overwrite each other. Tested on three more data sets (Gresik, Palembang, plus Bitung/Tomohon unchanged). 1.3.0 The FDT stays where your data puts it: a point whose name is an FDT code followed by more text (for example GSK1-CUL01-H03-D01-01, a distribution element) is never taken as the FDT, points that only look like an FDT by folder or attribute can no longer replace the point named with the FDT code, the FDT is written to the KMZ with its full original coordinates, and every result KMZ is opened again to confirm the FDT location equals the input; differences between the point, its LAT/LONG fields or the same FDT in the compilation are reported, never applied. Result KMZ follows the APD Feeder template exactly: each route folder now holds the Feeder folder with every template folder in template order (checked after writing), cable colours and line width follow the template (12C, 24C and width 3 corrected), icon sizes follow the template, and your organisation's own icons are used when you put them in the Tembuku "ikon" folder (they stay on your computer). Feeder tube records: an Excel/CSV table with Feeder and Used Tube columns is recognised and combined with the compilation; T numbers never go past 12 - an unrecorded number is proposed with a note to verify, and a feeder whose 12 numbers are all recorded is flagged. Cluster ID and area are read from file names like "<ID>_<area>.kmz"; route folder names written "APD Feeder_" are accepted without a note. Decisions for the 17 calibration FDTs are unchanged. 1.2.1 Progress you can see: while Tembuku works in the background, the Tembuku panel now shows a progress bar with the percentage and the current step (for example 'Designing feeders: 40 percent - Designing FDT FDT-12 (4/10)'), the percentage in the panel title, and a Cancel button; previously the progress was only in the small QGIS task bar at the bottom right. Applies to designing feeders, preparing data for map clicks, the accuracy test and clicked FDTs. No change in design results. 1.2.0 Plain language everywhere, so that people new to feeder design can follow every result: decisions and reasons are written as short sentences that start with the verdict; candidate feeders show "meets every condition", "usable, but ..." or "rejected: ..." with the reason in words (direction in degrees instead of a cosine, "route vs new feeder" instead of a ratio code, signal loss and what is left before the limit); input-check messages, the results panel, layer names and attribute fields in QGIS are fully in English, while the report, KMZ and CSV stay in Indonesian with readable column names and decision names instead of internal codes. Neutral vocabulary that fits any operator (cable inventory, feeder design compilation, KMZ with FDT points). Safety: an FDT that cannot be analysed is never dropped - it stays in every output as "Needs manual design" with the reason; text from KMZ files is fully encoded in the HTML report. Seventeen damaged or unusual input cases (empty, damaged or unsafe files, no POP, swapped coordinates, FDT outside Indonesia or 40 km from the POP, duplicate FDTs, odd cable sizes, two versions of a compilation) are tested and handled with a clear message. Smarter file recognition: every file is recognised from what is inside it, not from its name or extension - cable lines with fibre and spare-fibre data, folders holding both FDT points and cables (with or without F1_T1 names), lists of FDT/ODC points, POP/OLT/STO points - and the window and report say why. New formats: ZIP bundles, GeoJSON, CSV/TXT/Excel with coordinates (any common lat/lon spelling, WKT, decimal comma), and through QGIS Shapefile, GeoPackage, GML, TAB, FlatGeobuf and GPX; Indonesian and English column names and attributes kept in KMZ description tables are understood; damaged files are named clearly. Calculations and decisions are unchanged: the 17 test FDTs give exactly the same results as 1.1.1, and the same targets given as KMZ, CSV, GeoJSON or Excel give identical decisions. 1.1.1 Clearer plugin description and changelog; code-quality fix reported by the QGIS repository scan (E741). Decisions and results are unchanged. 1.1.0 Basemaps in QGIS and in the HTML report with current imagery only, because a picture older than the design hides new houses and roads: OpenStreetMap live; detailed Esri World Imagery (0.3-0.5 m in Indonesian towns, your own ArcGIS key) with the capture date at the map centre read from Esri's official metadata service and a warning when it is older than 12 months; the newest cloud-free Copernicus Sentinel-2 quarterly mosaic (10 m, your own free account; published within a month after each quarter, cloudy spots filled from earlier quarters, moves forward by itself). No dates are built in. Optical budget values updated to the maximum values of current ITU-T standards, each with its source (fibre 0.40 dB/km G.652 08/2024; splitters 1:4 7.4 dB and 1:8 10.5 dB G.9940 12/2023; connector 0.5 dB G.671 11/2025; splice 0.15 dB; class B+ 13-28 dB G.984.2); the 17 test decisions are unchanged. English interface with plain-language decision names, tooltips and an offline Help and Quick Guide with glossary. No company or brand names in the plugin: neutral labels, optional company labels in a local istilah.json. KMZ icons embedded in the file. HTML report redesigned with the Plus Jakarta Sans typeface (Tokotype, Jakarta; SIL OFL), a glossary and OpenStreetMap road lines stored for offline use. OpenStreetMap and Overpass use follows the server policies (application User-Agent, 30-60 s pause when the server is busy, own server configurable). New logo. Clearer error messages. 1.0.3 Road download retries when the public server is busy; a failed area no longer stops the others. 1.0.2 Accuracy test against the designs already in the compilation; feeder numbering per POP fixed. 1.0.1 Input checks follow ISO 19157-1:2023; fuzzy matching only for names, never numbers; digit-swapped cable sizes flagged. 1.0.0 First release.
